add updates libs

This commit is contained in:
Christoph Brandau
2026-06-18 17:21:58 +02:00
parent 91ce762570
commit 41f331d4c4
150 changed files with 8249 additions and 2550 deletions
+1 -1
View File
@@ -6,7 +6,7 @@ __title__ = "packaging"
__summary__ = "Core utilities for Python packages"
__uri__ = "https://github.com/pypa/packaging"
__version__ = "25.0"
__version__ = "26.2"
__author__ = "Donald Stufft and individual contributors"
__email__ = "donald@stufft.io"
-1
View File
@@ -4,7 +4,6 @@ ELF file parser.
This provides a class ``ELFFile`` that parses an ELF executable in a similar
interface to ``ZipFile``. Only the read interface is implemented.
Based on: https://gist.github.com/lyssdod/f51579ae8d93c8657a5564aefc2ffbca
ELF header: https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.eheader.html
"""
+36 -36
View File
@@ -15,6 +15,16 @@ EF_ARM_ABIMASK = 0xFF000000
EF_ARM_ABI_VER5 = 0x05000000
EF_ARM_ABI_FLOAT_HARD = 0x00000400
_ALLOWED_ARCHS = {
"x86_64",
"aarch64",
"ppc64",
"ppc64le",
"s390x",
"loongarch64",
"riscv64",
}
# `os.PathLike` not a generic type until Python 3.9, so sticking with `str`
# as the type for `path` until then.
@@ -57,16 +67,7 @@ def _have_compatible_abi(executable: str, archs: Sequence[str]) -> bool:
return _is_linux_armhf(executable)
if "i686" in archs:
return _is_linux_i686(executable)
allowed_archs = {
"x86_64",
"aarch64",
"ppc64",
"ppc64le",
"s390x",
"loongarch64",
"riscv64",
}
return any(arch in allowed_archs for arch in archs)
return any(arch in _ALLOWED_ARCHS for arch in archs)
# If glibc ever changes its major version, we need to know what the last
@@ -106,7 +107,7 @@ def _glibc_version_string_ctypes() -> str | None:
Fallback implementation of glibc_version_string using ctypes.
"""
try:
import ctypes
import ctypes # noqa: PLC0415
except ImportError:
return None
@@ -150,7 +151,7 @@ def _glibc_version_string() -> str | None:
return _glibc_version_string_confstr() or _glibc_version_string_ctypes()
def _parse_glibc_version(version_str: str) -> tuple[int, int]:
def _parse_glibc_version(version_str: str) -> _GLibCVersion:
"""Parse glibc version.
We use a regexp instead of str.split because we want to discard any
@@ -165,15 +166,15 @@ def _parse_glibc_version(version_str: str) -> tuple[int, int]:
RuntimeWarning,
stacklevel=2,
)
return -1, -1
return int(m.group("major")), int(m.group("minor"))
return _GLibCVersion(-1, -1)
return _GLibCVersion(int(m.group("major")), int(m.group("minor")))
@functools.lru_cache
def _get_glibc_version() -> tuple[int, int]:
def _get_glibc_version() -> _GLibCVersion:
version_str = _glibc_version_string()
if version_str is None:
return (-1, -1)
return _GLibCVersion(-1, -1)
return _parse_glibc_version(version_str)
@@ -184,7 +185,7 @@ def _is_compatible(arch: str, version: _GLibCVersion) -> bool:
return False
# Check for presence of _manylinux module.
try:
import _manylinux
import _manylinux # noqa: PLC0415
except ImportError:
return True
if hasattr(_manylinux, "manylinux_compatible"):
@@ -192,25 +193,26 @@ def _is_compatible(arch: str, version: _GLibCVersion) -> bool:
if result is not None:
return bool(result)
return True
if version == _GLibCVersion(2, 5):
if hasattr(_manylinux, "manylinux1_compatible"):
return bool(_manylinux.manylinux1_compatible)
if version == _GLibCVersion(2, 12):
if hasattr(_manylinux, "manylinux2010_compatible"):
return bool(_manylinux.manylinux2010_compatible)
if version == _GLibCVersion(2, 17):
if hasattr(_manylinux, "manylinux2014_compatible"):
return bool(_manylinux.manylinux2014_compatible)
if version == _GLibCVersion(2, 5) and hasattr(_manylinux, "manylinux1_compatible"):
return bool(_manylinux.manylinux1_compatible)
if version == _GLibCVersion(2, 12) and hasattr(
_manylinux, "manylinux2010_compatible"
):
return bool(_manylinux.manylinux2010_compatible)
if version == _GLibCVersion(2, 17) and hasattr(
_manylinux, "manylinux2014_compatible"
):
return bool(_manylinux.manylinux2014_compatible)
return True
_LEGACY_MANYLINUX_MAP = {
_LEGACY_MANYLINUX_MAP: dict[_GLibCVersion, str] = {
# CentOS 7 w/ glibc 2.17 (PEP 599)
(2, 17): "manylinux2014",
_GLibCVersion(2, 17): "manylinux2014",
# CentOS 6 w/ glibc 2.12 (PEP 571)
(2, 12): "manylinux2010",
_GLibCVersion(2, 12): "manylinux2010",
# CentOS 5 w/ glibc 2.5 (PEP 513)
(2, 5): "manylinux1",
_GLibCVersion(2, 5): "manylinux1",
}
@@ -252,11 +254,9 @@ def platform_tags(archs: Sequence[str]) -> Iterator[str]:
min_minor = -1
for glibc_minor in range(glibc_max.minor, min_minor, -1):
glibc_version = _GLibCVersion(glibc_max.major, glibc_minor)
tag = "manylinux_{}_{}".format(*glibc_version)
if _is_compatible(arch, glibc_version):
yield f"{tag}_{arch}"
# Handle the legacy manylinux1, manylinux2010, manylinux2014 tags.
if glibc_version in _LEGACY_MANYLINUX_MAP:
legacy_tag = _LEGACY_MANYLINUX_MAP[glibc_version]
if _is_compatible(arch, glibc_version):
yield "manylinux_{}_{}_{}".format(*glibc_version, arch)
# Handle the legacy manylinux1, manylinux2010, manylinux2014 tags.
if legacy_tag := _LEGACY_MANYLINUX_MAP.get(glibc_version):
yield f"{legacy_tag}_{arch}"
+1 -1
View File
@@ -49,7 +49,7 @@ def _get_musl_version(executable: str) -> _MuslVersion | None:
return None
if ld is None or "musl" not in ld:
return None
proc = subprocess.run([ld], stderr=subprocess.PIPE, text=True)
proc = subprocess.run([ld], check=False, stderr=subprocess.PIPE, text=True)
return _parse_musl_version(proc.stderr)
+50 -10
View File
@@ -7,12 +7,14 @@ the implementation.
from __future__ import annotations
import ast
from typing import NamedTuple, Sequence, Tuple, Union
from typing import List, Literal, NamedTuple, Sequence, Tuple, Union
from ._tokenizer import DEFAULT_RULES, Tokenizer
class Node:
__slots__ = ("value",)
def __init__(self, value: str) -> None:
self.value = value
@@ -20,31 +22,66 @@ class Node:
return self.value
def __repr__(self) -> str:
return f"<{self.__class__.__name__}('{self}')>"
return f"<{self.__class__.__name__}({self.value!r})>"
def serialize(self) -> str:
raise NotImplementedError
def __getstate__(self) -> str:
# Return just the value string for compactness and stability.
return self.value
def _restore_value(self, value: object) -> None:
if not isinstance(value, str):
raise TypeError(
f"Cannot restore {self.__class__.__name__} value from {value!r}"
)
self.value = value
def __setstate__(self, state: object) -> None:
if isinstance(state, str):
# New format (26.2+): just the value string.
self._restore_value(state)
return
if isinstance(state, tuple) and len(state) == 2:
# Old format (packaging <= 26.0, __slots__): (None, {slot: value}).
_, slot_dict = state
if isinstance(slot_dict, dict) and "value" in slot_dict:
self._restore_value(slot_dict["value"])
return
if isinstance(state, dict) and "value" in state:
# Old format (packaging <= 25.0, no __slots__): plain __dict__.
self._restore_value(state["value"])
return
raise TypeError(f"Cannot restore {self.__class__.__name__} from {state!r}")
class Variable(Node):
__slots__ = ()
def serialize(self) -> str:
return str(self)
class Value(Node):
__slots__ = ()
def serialize(self) -> str:
return f'"{self}"'
class Op(Node):
__slots__ = ()
def serialize(self) -> str:
return str(self)
MarkerLogical = Literal["and", "or"]
MarkerVar = Union[Variable, Value]
MarkerItem = Tuple[MarkerVar, Op, MarkerVar]
MarkerAtom = Union[MarkerItem, Sequence["MarkerAtom"]]
MarkerList = Sequence[Union["MarkerList", MarkerAtom, str]]
MarkerList = List[Union["MarkerList", MarkerAtom, MarkerLogical]]
class ParsedRequirement(NamedTuple):
@@ -111,7 +148,9 @@ def _parse_requirement_details(
return (url, specifier, marker)
marker = _parse_requirement_marker(
tokenizer, span_start=url_start, after="URL and whitespace"
tokenizer,
span_start=url_start,
expected="semicolon (after URL and whitespace)",
)
else:
specifier_start = tokenizer.position
@@ -124,10 +163,10 @@ def _parse_requirement_details(
marker = _parse_requirement_marker(
tokenizer,
span_start=specifier_start,
after=(
"version specifier"
expected=(
"comma (within version specifier), semicolon (after version specifier)"
if specifier
else "name and no valid version specifier"
else "semicolon (after name with no version specifier)"
),
)
@@ -135,7 +174,7 @@ def _parse_requirement_details(
def _parse_requirement_marker(
tokenizer: Tokenizer, *, span_start: int, after: str
tokenizer: Tokenizer, *, span_start: int, expected: str
) -> MarkerList:
"""
requirement_marker = SEMICOLON marker WS?
@@ -143,8 +182,9 @@ def _parse_requirement_marker(
if not tokenizer.check("SEMICOLON"):
tokenizer.raise_syntax_error(
f"Expected end or semicolon (after {after})",
f"Expected {expected} or end",
span_start=span_start,
span_end=None,
)
tokenizer.read()
@@ -307,7 +347,7 @@ def _parse_marker_item(tokenizer: Tokenizer) -> MarkerItem:
return (marker_var_left, marker_op, marker_var_right)
def _parse_marker_var(tokenizer: Tokenizer) -> MarkerVar:
def _parse_marker_var(tokenizer: Tokenizer) -> MarkerVar: # noqa: RET503
"""
marker_var = VARIABLE | QUOTED_STRING
"""
+17 -45
View File
@@ -2,60 +2,32 @@
# 2.0, and the BSD License. See the LICENSE file in the root of this repository
# for complete details.
"""Backward-compatibility shim for unpickling Version objects serialized before
packaging 26.1.
Old pickles reference ``packaging._structures.InfinityType`` and
``packaging._structures.NegativeInfinityType``. This module provides minimal
stand-in classes so that ``pickle.loads()`` can resolve those references.
The deserialized objects are not used for comparisons — ``Version.__setstate__``
discards the stale ``_key`` cache and recomputes it from the core version fields.
"""
from __future__ import annotations
class InfinityType:
"""Stand-in for the removed ``InfinityType`` used in old comparison keys."""
def __repr__(self) -> str:
return "Infinity"
def __hash__(self) -> int:
return hash(repr(self))
def __lt__(self, other: object) -> bool:
return False
def __le__(self, other: object) -> bool:
return False
def __eq__(self, other: object) -> bool:
return isinstance(other, self.__class__)
def __gt__(self, other: object) -> bool:
return True
def __ge__(self, other: object) -> bool:
return True
def __neg__(self: object) -> "NegativeInfinityType":
return NegativeInfinity
Infinity = InfinityType()
class NegativeInfinityType:
"""Stand-in for the removed ``NegativeInfinityType`` used in old comparison keys."""
def __repr__(self) -> str:
return "-Infinity"
def __hash__(self) -> int:
return hash(repr(self))
def __lt__(self, other: object) -> bool:
return True
def __le__(self, other: object) -> bool:
return True
def __eq__(self, other: object) -> bool:
return isinstance(other, self.__class__)
def __gt__(self, other: object) -> bool:
return False
def __ge__(self, other: object) -> bool:
return False
def __neg__(self: object) -> InfinityType:
return Infinity
Infinity = InfinityType()
NegativeInfinity = NegativeInfinityType()
+24 -26
View File
@@ -3,7 +3,7 @@ from __future__ import annotations
import contextlib
import re
from dataclasses import dataclass
from typing import Iterator, NoReturn
from typing import Generator, Mapping, NoReturn
from .specifiers import Specifier
@@ -33,16 +33,16 @@ class ParserSyntaxError(Exception):
def __str__(self) -> str:
marker = " " * self.span[0] + "~" * (self.span[1] - self.span[0]) + "^"
return "\n ".join([self.message, self.source, marker])
return f"{self.message}\n {self.source}\n {marker}"
DEFAULT_RULES: dict[str, str | re.Pattern[str]] = {
"LEFT_PARENTHESIS": r"\(",
"RIGHT_PARENTHESIS": r"\)",
"LEFT_BRACKET": r"\[",
"RIGHT_BRACKET": r"\]",
"SEMICOLON": r";",
"COMMA": r",",
DEFAULT_RULES: dict[str, re.Pattern[str]] = {
"LEFT_PARENTHESIS": re.compile(r"\("),
"RIGHT_PARENTHESIS": re.compile(r"\)"),
"LEFT_BRACKET": re.compile(r"\["),
"RIGHT_BRACKET": re.compile(r"\]"),
"SEMICOLON": re.compile(r";"),
"COMMA": re.compile(r","),
"QUOTED_STRING": re.compile(
r"""
(
@@ -53,10 +53,10 @@ DEFAULT_RULES: dict[str, str | re.Pattern[str]] = {
""",
re.VERBOSE,
),
"OP": r"(===|==|~=|!=|<=|>=|<|>)",
"BOOLOP": r"\b(or|and)\b",
"IN": r"\bin\b",
"NOT": r"\bnot\b",
"OP": re.compile(r"(===|==|~=|!=|<=|>=|<|>)"),
"BOOLOP": re.compile(r"\b(or|and)\b"),
"IN": re.compile(r"\bin\b"),
"NOT": re.compile(r"\bnot\b"),
"VARIABLE": re.compile(
r"""
\b(
@@ -75,16 +75,16 @@ DEFAULT_RULES: dict[str, str | re.Pattern[str]] = {
re.VERBOSE,
),
"SPECIFIER": re.compile(
Specifier._operator_regex_str + Specifier._version_regex_str,
Specifier._specifier_regex_str,
re.VERBOSE | re.IGNORECASE,
),
"AT": r"\@",
"URL": r"[^ \t]+",
"IDENTIFIER": r"\b[a-zA-Z0-9][a-zA-Z0-9._-]*\b",
"VERSION_PREFIX_TRAIL": r"\.\*",
"VERSION_LOCAL_LABEL_TRAIL": r"\+[a-z0-9]+(?:[-_\.][a-z0-9]+)*",
"WS": r"[ \t]+",
"END": r"$",
"AT": re.compile(r"\@"),
"URL": re.compile(r"[^ \t]+"),
"IDENTIFIER": re.compile(r"\b[a-zA-Z0-9][a-zA-Z0-9._-]*\b"),
"VERSION_PREFIX_TRAIL": re.compile(r"\.\*"),
"VERSION_LOCAL_LABEL_TRAIL": re.compile(r"\+[a-z0-9]+(?:[-_\.][a-z0-9]+)*"),
"WS": re.compile(r"[ \t]+"),
"END": re.compile(r"$"),
}
@@ -99,12 +99,10 @@ class Tokenizer:
self,
source: str,
*,
rules: dict[str, str | re.Pattern[str]],
rules: Mapping[str, re.Pattern[str]],
) -> None:
self.source = source
self.rules: dict[str, re.Pattern[str]] = {
name: re.compile(pattern) for name, pattern in rules.items()
}
self.rules = rules
self.next_token: Token | None = None
self.position = 0
@@ -174,7 +172,7 @@ class Tokenizer:
@contextlib.contextmanager
def enclosing_tokens(
self, open_token: str, close_token: str, *, around: str
) -> Iterator[None]:
) -> Generator[None, None, None]:
if self.check(open_token):
open_position = self.position
self.read()
+302
View File
@@ -0,0 +1,302 @@
from __future__ import annotations
import re
from collections.abc import Mapping, Sequence
from .errors import _ErrorCollector
from .requirements import Requirement
__all__ = [
"CyclicDependencyGroup",
"DependencyGroupInclude",
"DependencyGroupResolver",
"DuplicateGroupNames",
"InvalidDependencyGroupObject",
"resolve_dependency_groups",
]
def __dir__() -> list[str]:
return __all__
# -----------
# Error Types
# -----------
class DuplicateGroupNames(ValueError):
"""
The same dependency groups were defined twice, with different non-normalized names.
"""
class CyclicDependencyGroup(ValueError):
"""
The dependency group includes form a cycle.
"""
def __init__(self, requested_group: str, group: str, include_group: str) -> None:
self.requested_group = requested_group
self.group = group
self.include_group = include_group
if include_group == group:
reason = f"{group} includes itself"
else:
reason = f"{include_group} -> {group}, {group} -> {include_group}"
super().__init__(
"Cyclic dependency group include while resolving "
f"{requested_group}: {reason}"
)
# in the PEP 735 spec, the tables in dependency group lists were described as
# "Dependency Object Specifiers", but the only defined type of object was a
# "Dependency Group Include" -- hence the naming of this error as "Object"
class InvalidDependencyGroupObject(ValueError):
"""
A member of a dependency group was identified as a dict, but was not in a valid
format.
"""
# ------------------------
# Object Model & Interface
# ------------------------
class DependencyGroupInclude:
__slots__ = ("include_group",)
def __init__(self, include_group: str) -> None:
"""
Initialize a DependencyGroupInclude.
:param include_group: The name of the group referred to by this include.
"""
self.include_group = include_group
def __repr__(self) -> str:
return f"{self.__class__.__name__}({self.include_group!r})"
class DependencyGroupResolver:
"""
A resolver for Dependency Group data.
This class handles caching, name normalization, cycle detection, and other
parsing requirements. There are only two public methods for exploring the data:
``lookup()`` and ``resolve()``.
:param dependency_groups: A mapping, as provided via pyproject
``[dependency-groups]``.
"""
def __init__(
self,
dependency_groups: Mapping[str, Sequence[str | Mapping[str, str]]],
) -> None:
errors = _ErrorCollector()
self.dependency_groups = _normalize_group_names(dependency_groups, errors)
# a map of group names to parsed data
self._parsed_groups: dict[
str, tuple[Requirement | DependencyGroupInclude, ...]
] = {}
# a map of group names to their ancestors, used for cycle detection
self._include_graph_ancestors: dict[str, tuple[str, ...]] = {}
# a cache of completed resolutions to Requirement lists
self._resolve_cache: dict[str, tuple[Requirement, ...]] = {}
errors.finalize("[dependency-groups] data was invalid")
def lookup(self, group: str) -> tuple[Requirement | DependencyGroupInclude, ...]:
"""
Lookup a group name, returning the parsed dependency data for that group.
This will not resolve includes.
:param group: the name of the group to lookup
"""
group = _normalize_name(group)
with _ErrorCollector().on_exit(
f"[dependency-groups] data for {group!r} was malformed"
) as errors:
return self._parse_group(group, errors)
def resolve(self, group: str) -> tuple[Requirement, ...]:
"""
Resolve a dependency group to a list of requirements.
:param group: the name of the group to resolve
"""
group = _normalize_name(group)
with _ErrorCollector().on_exit(
f"[dependency-groups] data for {group!r} was malformed"
) as errors:
return self._resolve(group, group, errors)
def _resolve(
self, group: str, requested_group: str, errors: _ErrorCollector
) -> tuple[Requirement, ...]:
"""
This is a helper for cached resolution to strings. It preserves the name of the
group which the user initially requested in order to present a clearer error in
the event that a cycle is detected.
:param group: The normalized name of the group to resolve.
:param requested_group: The group which was used in the original, user-facing
request.
"""
if group in self._resolve_cache:
return self._resolve_cache[group]
parsed = self._parse_group(group, errors)
resolved_group = []
for item in parsed:
if isinstance(item, Requirement):
resolved_group.append(item)
elif isinstance(item, DependencyGroupInclude):
include_group = _normalize_name(item.include_group)
# if a group is cyclic, record the error
# otherwise, follow the include_group reference
#
# this allows us to examine all includes in a group, even in the
# presence of errors
if include_group in self._include_graph_ancestors.get(group, ()):
errors.error(
CyclicDependencyGroup(
requested_group, group, item.include_group
)
)
else:
self._include_graph_ancestors[include_group] = (
*self._include_graph_ancestors.get(group, ()),
group,
)
resolved_group.extend(
self._resolve(include_group, requested_group, errors)
)
else: # pragma: no cover
raise NotImplementedError(
f"Invalid dependency group item after parse: {item}"
)
# in the event that errors were detected, present the group as empty and do not
# cache the result
# this ensures that repeated access to a cyclic group will raise multiple errors
if errors.errors:
return ()
self._resolve_cache[group] = tuple(resolved_group)
return self._resolve_cache[group]
def _parse_group(
self, group: str, errors: _ErrorCollector
) -> tuple[Requirement | DependencyGroupInclude, ...]:
# short circuit -- never do the work twice
if group in self._parsed_groups:
return self._parsed_groups[group]
if group not in self.dependency_groups:
errors.error(LookupError(f"Dependency group '{group}' not found"))
return ()
raw_group = self.dependency_groups[group]
if isinstance(raw_group, str):
errors.error(
TypeError(
f"Dependency group {group!r} contained a string rather than a list."
)
)
return ()
if not isinstance(raw_group, Sequence):
errors.error(
TypeError(f"Dependency group {group!r} is not a sequence type.")
)
return ()
elements: list[Requirement | DependencyGroupInclude] = []
for item in raw_group:
if isinstance(item, str):
# packaging.requirements.Requirement parsing ensures that this is a
# valid PEP 508 Dependency Specifier
# raises InvalidRequirement on failure
elements.append(Requirement(item))
elif isinstance(item, Mapping):
if tuple(item.keys()) != ("include-group",):
errors.error(
InvalidDependencyGroupObject(
f"Invalid dependency group item: {item!r}"
)
)
else:
include_group = item["include-group"]
elements.append(DependencyGroupInclude(include_group=include_group))
else:
errors.error(TypeError(f"Invalid dependency group item: {item!r}"))
self._parsed_groups[group] = tuple(elements)
return self._parsed_groups[group]
# --------------------
# Functional Interface
# --------------------
def resolve_dependency_groups(
dependency_groups: Mapping[str, Sequence[str | Mapping[str, str]]], /, *groups: str
) -> tuple[str, ...]:
"""
Resolve a dependency group to a tuple of requirements, as strings.
:param dependency_groups: the parsed contents of the ``[dependency-groups]`` table
from ``pyproject.toml``
:param groups: the name of the group(s) to resolve
"""
resolver = DependencyGroupResolver(dependency_groups)
return tuple(str(r) for group in groups for r in resolver.resolve(group))
# ----------------
# internal helpers
# ----------------
_NORMALIZE_PATTERN = re.compile(r"[-_.]+")
def _normalize_name(name: str) -> str:
return _NORMALIZE_PATTERN.sub("-", name).lower()
def _normalize_group_names(
dependency_groups: Mapping[str, Sequence[str | Mapping[str, str]]],
errors: _ErrorCollector,
) -> dict[str, Sequence[str | Mapping[str, str]]]:
original_names: dict[str, list[str]] = {}
normalized_groups: dict[str, Sequence[str | Mapping[str, str]]] = {}
for group_name, value in dependency_groups.items():
normed_group_name = _normalize_name(group_name)
original_names.setdefault(normed_group_name, []).append(group_name)
normalized_groups[normed_group_name] = value
for normed_name, names in original_names.items():
if len(names) > 1:
errors.error(
DuplicateGroupNames(
"Duplicate dependency group names: "
f"{normed_name} ({', '.join(names)})"
)
)
return normalized_groups
+325
View File
@@ -0,0 +1,325 @@
from __future__ import annotations
import dataclasses
import re
import urllib.parse
from collections.abc import Mapping
from typing import TYPE_CHECKING, Any, Protocol, TypeVar
if TYPE_CHECKING: # pragma: no cover
import sys
from collections.abc import Collection
if sys.version_info >= (3, 11):
from typing import Self
else:
from typing_extensions import Self
__all__ = [
"ArchiveInfo",
"DirInfo",
"DirectUrl",
"DirectUrlValidationError",
"VcsInfo",
]
def __dir__() -> list[str]:
return __all__
_T = TypeVar("_T")
class _FromMappingProtocol(Protocol): # pragma: no cover
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self: ...
_FromMappingProtocolT = TypeVar("_FromMappingProtocolT", bound=_FromMappingProtocol)
def _json_dict_factory(data: list[tuple[str, Any]]) -> dict[str, Any]:
return {key: value for key, value in data if value is not None}
def _get(d: Mapping[str, Any], expected_type: type[_T], key: str) -> _T | None:
"""Get a value from the dictionary and verify it's the expected type."""
if (value := d.get(key)) is None:
return None
if not isinstance(value, expected_type):
raise DirectUrlValidationError(
f"Unexpected type {type(value).__name__} "
f"(expected {expected_type.__name__})",
context=key,
)
return value
def _get_required(d: Mapping[str, Any], expected_type: type[_T], key: str) -> _T:
"""Get a required value from the dictionary and verify it's the expected type."""
if (value := _get(d, expected_type, key)) is None:
raise _DirectUrlRequiredKeyError(key)
return value
def _get_object(
d: Mapping[str, Any], target_type: type[_FromMappingProtocolT], key: str
) -> _FromMappingProtocolT | None:
"""Get a dictionary value from the dictionary and convert it to a dataclass."""
if (value := _get(d, Mapping, key)) is None: # type: ignore[type-abstract]
return None
try:
return target_type._from_dict(value)
except Exception as e:
raise DirectUrlValidationError(e, context=key) from e
_PEP610_USER_PASS_ENV_VARS_REGEX = re.compile(
r"^\$\{[A-Za-z0-9-_]+\}(:\$\{[A-Za-z0-9-_]+\})?$"
)
def _strip_auth_from_netloc(netloc: str, safe_user_passwords: Collection[str]) -> str:
if "@" not in netloc:
return netloc
user_pass, netloc_no_user_pass = netloc.split("@", 1)
if user_pass in safe_user_passwords:
return netloc
if _PEP610_USER_PASS_ENV_VARS_REGEX.match(user_pass):
return netloc
return netloc_no_user_pass
def _strip_url(url: str, safe_user_passwords: Collection[str]) -> str:
"""url with user:password part removed unless it is formed with
environment variables as specified in PEP 610, or it is a safe user:password
such as `git`.
"""
parsed_url = urllib.parse.urlsplit(url)
netloc = _strip_auth_from_netloc(parsed_url.netloc, safe_user_passwords)
return urllib.parse.urlunsplit(
(
parsed_url.scheme,
netloc,
parsed_url.path,
parsed_url.query,
parsed_url.fragment,
)
)
class DirectUrlValidationError(Exception):
"""Raised when when input data is not spec-compliant."""
context: str | None = None
message: str
def __init__(
self,
cause: str | Exception,
*,
context: str | None = None,
) -> None:
if isinstance(cause, DirectUrlValidationError):
if cause.context:
self.context = (
f"{context}.{cause.context}" if context else cause.context
)
else:
self.context = context # pragma: no cover
self.message = cause.message
else:
self.context = context
self.message = str(cause)
def __str__(self) -> str:
if self.context:
return f"{self.message} in {self.context!r}"
return self.message
class _DirectUrlRequiredKeyError(DirectUrlValidationError):
def __init__(self, key: str) -> None:
super().__init__("Missing required value", context=key)
@dataclasses.dataclass(frozen=True, init=False)
class VcsInfo:
vcs: str
commit_id: str
requested_revision: str | None = None
def __init__(
self,
*,
vcs: str,
commit_id: str,
requested_revision: str | None = None,
) -> None:
object.__setattr__(self, "vcs", vcs)
object.__setattr__(self, "commit_id", commit_id)
object.__setattr__(self, "requested_revision", requested_revision)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
# We can't validate vcs value because is not closed.
return cls(
vcs=_get_required(d, str, "vcs"),
requested_revision=_get(d, str, "requested_revision"),
commit_id=_get_required(d, str, "commit_id"),
)
@dataclasses.dataclass(frozen=True, init=False)
class ArchiveInfo:
hashes: Mapping[str, str] | None = None
def __init__(
self,
*,
hashes: Mapping[str, str] | None = None,
) -> None:
object.__setattr__(self, "hashes", hashes)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
hashes = _get(d, Mapping, "hashes") # type: ignore[type-abstract]
if hashes is not None and not all(isinstance(h, str) for h in hashes.values()):
raise DirectUrlValidationError(
"Hash values must be strings", context="hashes"
)
legacy_hash = _get(d, str, "hash")
if legacy_hash is not None:
if "=" not in legacy_hash:
raise DirectUrlValidationError(
"Invalid hash format (expected '<algorithm>=<hash>')",
context="hash",
)
hash_algorithm, hash_value = legacy_hash.split("=", 1)
if hashes is None:
# if `hashes` are not present, we can derive it from the legacy `hash`
hashes = {hash_algorithm: hash_value}
else:
# if `hashes` are present, the legacy `hash` must match one of them
if hash_algorithm not in hashes:
raise DirectUrlValidationError(
f"Algorithm {hash_algorithm!r} used in hash field "
f"is not present in hashes field",
context="hashes",
)
if hashes[hash_algorithm] != hash_value:
raise DirectUrlValidationError(
f"Algorithm {hash_algorithm!r} used in hash field "
f"has different value in hashes field",
context="hash",
)
return cls(hashes=hashes)
@dataclasses.dataclass(frozen=True, init=False)
class DirInfo:
editable: bool | None = None
def __init__(
self,
*,
editable: bool | None = None,
) -> None:
object.__setattr__(self, "editable", editable)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
return cls(
editable=_get(d, bool, "editable"),
)
@dataclasses.dataclass(frozen=True, init=False)
class DirectUrl:
"""A class representing a direct URL."""
url: str
archive_info: ArchiveInfo | None = None
vcs_info: VcsInfo | None = None
dir_info: DirInfo | None = None
subdirectory: str | None = None # XXX Path or str?
def __init__(
self,
*,
url: str,
archive_info: ArchiveInfo | None = None,
vcs_info: VcsInfo | None = None,
dir_info: DirInfo | None = None,
subdirectory: str | None = None,
) -> None:
object.__setattr__(self, "url", url)
object.__setattr__(self, "archive_info", archive_info)
object.__setattr__(self, "vcs_info", vcs_info)
object.__setattr__(self, "dir_info", dir_info)
object.__setattr__(self, "subdirectory", subdirectory)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
direct_url = cls(
url=_get_required(d, str, "url"),
archive_info=_get_object(d, ArchiveInfo, "archive_info"),
vcs_info=_get_object(d, VcsInfo, "vcs_info"),
dir_info=_get_object(d, DirInfo, "dir_info"),
subdirectory=_get(d, str, "subdirectory"),
)
if (
bool(direct_url.vcs_info)
+ bool(direct_url.archive_info)
+ bool(direct_url.dir_info)
) != 1:
raise DirectUrlValidationError(
"Exactly one of vcs_info, archive_info, dir_info must be present"
)
if direct_url.dir_info is not None and not direct_url.url.startswith("file://"):
raise DirectUrlValidationError(
"URL scheme must be file:// when dir_info is present",
context="url",
)
# XXX subdirectory must be relative, can we, should we validate that here?
return direct_url
@classmethod
def from_dict(cls, d: Mapping[str, Any], /) -> Self:
"""Create and validate a DirectUrl instance from a JSON dictionary."""
return cls._from_dict(d)
def to_dict(
self,
*,
generate_legacy_hash: bool = False,
strip_user_password: bool = True,
safe_user_passwords: Collection[str] = ("git",),
) -> Mapping[str, Any]:
"""Convert the DirectUrl instance to a JSON dictionary.
:param generate_legacy_hash: If True, include a legacy `hash` field in
`archive_info` for backward compatibility with tools that don't
support the `hashes` field.
:param strip_user_password: If True, strip user:password from the URL
unless it is formed with environment variables as specified in PEP
610, or it is a safe user:password such as `git`.
:param safe_user_passwords: A collection of user:password strings that
should not be stripped from the URL even if `strip_user_password` is
True.
"""
res = dataclasses.asdict(self, dict_factory=_json_dict_factory)
if generate_legacy_hash and self.archive_info and self.archive_info.hashes:
hash_algorithm, hash_value = next(iter(self.archive_info.hashes.items()))
res["archive_info"]["hash"] = f"{hash_algorithm}={hash_value}"
if strip_user_password:
res["url"] = _strip_url(self.url, safe_user_passwords)
return res
def validate(self) -> None:
"""Validate the DirectUrl instance against the specification.
Raises :class:`DirectUrlValidationError` if invalid.
"""
self.from_dict(self.to_dict())
+94
View File
@@ -0,0 +1,94 @@
from __future__ import annotations
import contextlib
import dataclasses
import sys
import typing
__all__ = ["ExceptionGroup"]
def __dir__() -> list[str]:
return __all__
if sys.version_info >= (3, 11): # pragma: no cover
from builtins import ExceptionGroup
else: # pragma: no cover
class ExceptionGroup(Exception):
"""A minimal implementation of :external:exc:`ExceptionGroup` from Python 3.11.
If :external:exc:`ExceptionGroup` is already defined by Python itself,
that version is used instead.
"""
message: str
exceptions: list[Exception]
def __init__(self, message: str, exceptions: list[Exception]) -> None:
self.message = message
self.exceptions = exceptions
def __repr__(self) -> str:
return f"{self.__class__.__name__}({self.message!r}, {self.exceptions!r})"
@dataclasses.dataclass
class _ErrorCollector:
"""
Collect errors into ExceptionGroups.
Used like this:
collector = _ErrorCollector()
# Add a single exception
collector.error(ValueError("one"))
# Supports nesting, including combining ExceptionGroups
with collector.collect():
raise ValueError("two")
collector.finalize("Found some errors")
Since making a collector and then calling finalize later is a common pattern,
a convenience method ``on_exit`` is provided.
"""
errors: list[Exception] = dataclasses.field(default_factory=list, init=False)
def finalize(self, msg: str) -> None:
"""Raise a group exception if there are any errors."""
if self.errors:
raise ExceptionGroup(msg, self.errors)
@contextlib.contextmanager
def on_exit(self, msg: str) -> typing.Generator[_ErrorCollector, None, None]:
"""
Calls finalize if no uncollected errors were present.
Uncollected errors are raised normally.
"""
yield self
self.finalize(msg)
@contextlib.contextmanager
def collect(self, *err_cls: type[Exception]) -> typing.Generator[None, None, None]:
"""
Context manager to collect errors into the error list.
Must be inside loops, as only one error can be collected at a time.
"""
error_classes = err_cls or (Exception,)
try:
yield
except ExceptionGroup as error:
self.errors.extend(error.exceptions)
except error_classes as error:
self.errors.append(error)
def error(
self,
error: Exception,
) -> None:
"""Add an error to the list."""
self.errors.append(error)
+52 -11
View File
@@ -34,7 +34,7 @@ from __future__ import annotations
import re
from typing import NewType, cast
from packaging.licenses._spdx import EXCEPTIONS, LICENSES
from ._spdx import EXCEPTIONS, LICENSES
__all__ = [
"InvalidLicenseExpression",
@@ -42,14 +42,25 @@ __all__ = [
"canonicalize_license_expression",
]
# Simple __dir__ implementation since there are no public submodules
def __dir__() -> list[str]:
return __all__
license_ref_allowed = re.compile("^[A-Za-z0-9.-]*$")
NormalizedLicenseExpression = NewType("NormalizedLicenseExpression", str)
"""
A :class:`typing.NewType` of :class:`str`, representing a normalized
License-Expression.
"""
class InvalidLicenseExpression(ValueError):
"""Raised when a license-expression string is invalid
>>> from packaging.licenses import canonicalize_license_expression
>>> canonicalize_license_expression("invalid")
Traceback (most recent call last):
...
@@ -60,6 +71,34 @@ class InvalidLicenseExpression(ValueError):
def canonicalize_license_expression(
raw_license_expression: str,
) -> NormalizedLicenseExpression:
"""
This function takes a valid License-Expression, and returns the normalized
form of it.
The return type is typed as :class:`NormalizedLicenseExpression`. This
allows type checkers to help require that a string has passed through this
function before use.
:param str raw_license_expression: The License-Expression to canonicalize.
:raises InvalidLicenseExpression: If the License-Expression is invalid due to an
invalid/unknown license identifier or invalid syntax.
.. doctest::
>>> from packaging.licenses import canonicalize_license_expression
>>> canonicalize_license_expression("mit")
'MIT'
>>> canonicalize_license_expression("mit and (apache-2.0 or bsd-2-clause)")
'MIT AND (Apache-2.0 OR BSD-2-Clause)'
>>> canonicalize_license_expression("(mit")
Traceback (most recent call last):
...
InvalidLicenseExpression: Invalid license expression: '(mit'
>>> canonicalize_license_expression("Use-it-after-midnight")
Traceback (most recent call last):
...
InvalidLicenseExpression: Unknown license: 'Use-it-after-midnight'
"""
if not raw_license_expression:
message = f"Invalid license expression: {raw_license_expression!r}"
raise InvalidLicenseExpression(message)
@@ -80,16 +119,21 @@ def canonicalize_license_expression(
tokens = license_expression.split()
# Rather than implementing boolean logic, we create an expression that Python can
# parse. Everything that is not involved with the grammar itself is treated as
# `False` and the expression should evaluate as such.
# Rather than implementing a parenthesis/boolean logic parser, create an
# expression that Python can parse. Everything that is not involved with the
# grammar itself is replaced with the placeholder `False` and the resultant
# expression should become a valid Python expression.
python_tokens = []
for token in tokens:
if token not in {"or", "and", "with", "(", ")"}:
python_tokens.append("False")
elif token == "with":
python_tokens.append("or")
elif token == "(" and python_tokens and python_tokens[-1] not in {"or", "and"}:
elif (
token == "("
and python_tokens
and python_tokens[-1] not in {"or", "and", "("}
) or (token == ")" and python_tokens and python_tokens[-1] == "("):
message = f"Invalid license expression: {raw_license_expression!r}"
raise InvalidLicenseExpression(message)
else:
@@ -97,11 +141,8 @@ def canonicalize_license_expression(
python_expression = " ".join(python_tokens)
try:
invalid = eval(python_expression, globals(), locals())
except Exception:
invalid = True
if invalid is not False:
compile(python_expression, "", "eval")
except SyntaxError:
message = f"Invalid license expression: {raw_license_expression!r}"
raise InvalidLicenseExpression(message) from None
@@ -140,6 +181,6 @@ def canonicalize_license_expression(
normalized_expression = " ".join(normalized_tokens)
return cast(
NormalizedLicenseExpression,
"NormalizedLicenseExpression",
normalized_expression.replace("( ", "(").replace(" )", ")"),
)
+41 -1
View File
@@ -12,7 +12,7 @@ class SPDXException(TypedDict):
deprecated: bool
VERSION = '3.25.0'
VERSION = '3.27.0'
LICENSES: dict[str, SPDXLicense] = {
'0bsd': {'id': '0BSD', 'deprecated': False},
@@ -46,6 +46,7 @@ LICENSES: dict[str, SPDXLicense] = {
'antlr-pd': {'id': 'ANTLR-PD', 'deprecated': False},
'antlr-pd-fallback': {'id': 'ANTLR-PD-fallback', 'deprecated': False},
'any-osi': {'id': 'any-OSI', 'deprecated': False},
'any-osi-perl-modules': {'id': 'any-OSI-perl-modules', 'deprecated': False},
'apache-1.0': {'id': 'Apache-1.0', 'deprecated': False},
'apache-1.1': {'id': 'Apache-1.1', 'deprecated': False},
'apache-2.0': {'id': 'Apache-2.0', 'deprecated': False},
@@ -61,6 +62,8 @@ LICENSES: dict[str, SPDXLicense] = {
'artistic-1.0-cl8': {'id': 'Artistic-1.0-cl8', 'deprecated': False},
'artistic-1.0-perl': {'id': 'Artistic-1.0-Perl', 'deprecated': False},
'artistic-2.0': {'id': 'Artistic-2.0', 'deprecated': False},
'artistic-dist': {'id': 'Artistic-dist', 'deprecated': False},
'aspell-ru': {'id': 'Aspell-RU', 'deprecated': False},
'aswf-digital-assets-1.0': {'id': 'ASWF-Digital-Assets-1.0', 'deprecated': False},
'aswf-digital-assets-1.1': {'id': 'ASWF-Digital-Assets-1.1', 'deprecated': False},
'baekmuk': {'id': 'Baekmuk', 'deprecated': False},
@@ -75,6 +78,7 @@ LICENSES: dict[str, SPDXLicense] = {
'blessing': {'id': 'blessing', 'deprecated': False},
'blueoak-1.0.0': {'id': 'BlueOak-1.0.0', 'deprecated': False},
'boehm-gc': {'id': 'Boehm-GC', 'deprecated': False},
'boehm-gc-without-fee': {'id': 'Boehm-GC-without-fee', 'deprecated': False},
'borceux': {'id': 'Borceux', 'deprecated': False},
'brian-gladman-2-clause': {'id': 'Brian-Gladman-2-Clause', 'deprecated': False},
'brian-gladman-3-clause': {'id': 'Brian-Gladman-3-Clause', 'deprecated': False},
@@ -85,6 +89,7 @@ LICENSES: dict[str, SPDXLicense] = {
'bsd-2-clause-freebsd': {'id': 'BSD-2-Clause-FreeBSD', 'deprecated': True},
'bsd-2-clause-netbsd': {'id': 'BSD-2-Clause-NetBSD', 'deprecated': True},
'bsd-2-clause-patent': {'id': 'BSD-2-Clause-Patent', 'deprecated': False},
'bsd-2-clause-pkgconf-disclaimer': {'id': 'BSD-2-Clause-pkgconf-disclaimer', 'deprecated': False},
'bsd-2-clause-views': {'id': 'BSD-2-Clause-Views', 'deprecated': False},
'bsd-3-clause': {'id': 'BSD-3-Clause', 'deprecated': False},
'bsd-3-clause-acpica': {'id': 'BSD-3-Clause-acpica', 'deprecated': False},
@@ -176,6 +181,8 @@ LICENSES: dict[str, SPDXLicense] = {
'cc-by-sa-3.0-igo': {'id': 'CC-BY-SA-3.0-IGO', 'deprecated': False},
'cc-by-sa-4.0': {'id': 'CC-BY-SA-4.0', 'deprecated': False},
'cc-pddc': {'id': 'CC-PDDC', 'deprecated': False},
'cc-pdm-1.0': {'id': 'CC-PDM-1.0', 'deprecated': False},
'cc-sa-1.0': {'id': 'CC-SA-1.0', 'deprecated': False},
'cc0-1.0': {'id': 'CC0-1.0', 'deprecated': False},
'cddl-1.0': {'id': 'CDDL-1.0', 'deprecated': False},
'cddl-1.1': {'id': 'CDDL-1.1', 'deprecated': False},
@@ -215,6 +222,7 @@ LICENSES: dict[str, SPDXLicense] = {
'cpol-1.02': {'id': 'CPOL-1.02', 'deprecated': False},
'cronyx': {'id': 'Cronyx', 'deprecated': False},
'crossword': {'id': 'Crossword', 'deprecated': False},
'cryptoswift': {'id': 'CryptoSwift', 'deprecated': False},
'crystalstacker': {'id': 'CrystalStacker', 'deprecated': False},
'cua-opl-1.0': {'id': 'CUA-OPL-1.0', 'deprecated': False},
'cube': {'id': 'Cube', 'deprecated': False},
@@ -226,7 +234,9 @@ LICENSES: dict[str, SPDXLicense] = {
'dl-de-by-2.0': {'id': 'DL-DE-BY-2.0', 'deprecated': False},
'dl-de-zero-2.0': {'id': 'DL-DE-ZERO-2.0', 'deprecated': False},
'doc': {'id': 'DOC', 'deprecated': False},
'docbook-dtd': {'id': 'DocBook-DTD', 'deprecated': False},
'docbook-schema': {'id': 'DocBook-Schema', 'deprecated': False},
'docbook-stylesheet': {'id': 'DocBook-Stylesheet', 'deprecated': False},
'docbook-xml': {'id': 'DocBook-XML', 'deprecated': False},
'dotseqn': {'id': 'Dotseqn', 'deprecated': False},
'drl-1.0': {'id': 'DRL-1.0', 'deprecated': False},
@@ -263,12 +273,17 @@ LICENSES: dict[str, SPDXLicense] = {
'fsfap-no-warranty-disclaimer': {'id': 'FSFAP-no-warranty-disclaimer', 'deprecated': False},
'fsful': {'id': 'FSFUL', 'deprecated': False},
'fsfullr': {'id': 'FSFULLR', 'deprecated': False},
'fsfullrsd': {'id': 'FSFULLRSD', 'deprecated': False},
'fsfullrwd': {'id': 'FSFULLRWD', 'deprecated': False},
'fsl-1.1-alv2': {'id': 'FSL-1.1-ALv2', 'deprecated': False},
'fsl-1.1-mit': {'id': 'FSL-1.1-MIT', 'deprecated': False},
'ftl': {'id': 'FTL', 'deprecated': False},
'furuseth': {'id': 'Furuseth', 'deprecated': False},
'fwlw': {'id': 'fwlw', 'deprecated': False},
'game-programming-gems': {'id': 'Game-Programming-Gems', 'deprecated': False},
'gcr-docs': {'id': 'GCR-docs', 'deprecated': False},
'gd': {'id': 'GD', 'deprecated': False},
'generic-xts': {'id': 'generic-xts', 'deprecated': False},
'gfdl-1.1': {'id': 'GFDL-1.1', 'deprecated': True},
'gfdl-1.1-invariants-only': {'id': 'GFDL-1.1-invariants-only', 'deprecated': False},
'gfdl-1.1-invariants-or-later': {'id': 'GFDL-1.1-invariants-or-later', 'deprecated': False},
@@ -320,6 +335,7 @@ LICENSES: dict[str, SPDXLicense] = {
'gtkbook': {'id': 'gtkbook', 'deprecated': False},
'gutmann': {'id': 'Gutmann', 'deprecated': False},
'haskellreport': {'id': 'HaskellReport', 'deprecated': False},
'hdf5': {'id': 'HDF5', 'deprecated': False},
'hdparm': {'id': 'hdparm', 'deprecated': False},
'hidapi': {'id': 'HIDAPI', 'deprecated': False},
'hippocratic-2.1': {'id': 'Hippocratic-2.1', 'deprecated': False},
@@ -360,6 +376,7 @@ LICENSES: dict[str, SPDXLicense] = {
'imlib2': {'id': 'Imlib2', 'deprecated': False},
'info-zip': {'id': 'Info-ZIP', 'deprecated': False},
'inner-net-2.0': {'id': 'Inner-Net-2.0', 'deprecated': False},
'innosetup': {'id': 'InnoSetup', 'deprecated': False},
'intel': {'id': 'Intel', 'deprecated': False},
'intel-acpi': {'id': 'Intel-ACPI', 'deprecated': False},
'interbase-1.0': {'id': 'Interbase-1.0', 'deprecated': False},
@@ -369,6 +386,7 @@ LICENSES: dict[str, SPDXLicense] = {
'isc-veillard': {'id': 'ISC-Veillard', 'deprecated': False},
'jam': {'id': 'Jam', 'deprecated': False},
'jasper-2.0': {'id': 'JasPer-2.0', 'deprecated': False},
'jove': {'id': 'jove', 'deprecated': False},
'jpl-image': {'id': 'JPL-image', 'deprecated': False},
'jpnic': {'id': 'JPNIC', 'deprecated': False},
'json': {'id': 'JSON', 'deprecated': False},
@@ -394,6 +412,7 @@ LICENSES: dict[str, SPDXLicense] = {
'lgpl-3.0-or-later': {'id': 'LGPL-3.0-or-later', 'deprecated': False},
'lgpllr': {'id': 'LGPLLR', 'deprecated': False},
'libpng': {'id': 'Libpng', 'deprecated': False},
'libpng-1.6.35': {'id': 'libpng-1.6.35', 'deprecated': False},
'libpng-2.0': {'id': 'libpng-2.0', 'deprecated': False},
'libselinux-1.0': {'id': 'libselinux-1.0', 'deprecated': False},
'libtiff': {'id': 'libtiff', 'deprecated': False},
@@ -424,14 +443,17 @@ LICENSES: dict[str, SPDXLicense] = {
'magaz': {'id': 'magaz', 'deprecated': False},
'mailprio': {'id': 'mailprio', 'deprecated': False},
'makeindex': {'id': 'MakeIndex', 'deprecated': False},
'man2html': {'id': 'man2html', 'deprecated': False},
'martin-birgmeier': {'id': 'Martin-Birgmeier', 'deprecated': False},
'mcphee-slideshow': {'id': 'McPhee-slideshow', 'deprecated': False},
'metamail': {'id': 'metamail', 'deprecated': False},
'minpack': {'id': 'Minpack', 'deprecated': False},
'mips': {'id': 'MIPS', 'deprecated': False},
'miros': {'id': 'MirOS', 'deprecated': False},
'mit': {'id': 'MIT', 'deprecated': False},
'mit-0': {'id': 'MIT-0', 'deprecated': False},
'mit-advertising': {'id': 'MIT-advertising', 'deprecated': False},
'mit-click': {'id': 'MIT-Click', 'deprecated': False},
'mit-cmu': {'id': 'MIT-CMU', 'deprecated': False},
'mit-enna': {'id': 'MIT-enna', 'deprecated': False},
'mit-feh': {'id': 'MIT-feh', 'deprecated': False},
@@ -472,6 +494,7 @@ LICENSES: dict[str, SPDXLicense] = {
'netcdf': {'id': 'NetCDF', 'deprecated': False},
'newsletr': {'id': 'Newsletr', 'deprecated': False},
'ngpl': {'id': 'NGPL', 'deprecated': False},
'ngrep': {'id': 'ngrep', 'deprecated': False},
'nicta-1.0': {'id': 'NICTA-1.0', 'deprecated': False},
'nist-pd': {'id': 'NIST-PD', 'deprecated': False},
'nist-pd-fallback': {'id': 'NIST-PD-fallback', 'deprecated': False},
@@ -486,6 +509,7 @@ LICENSES: dict[str, SPDXLicense] = {
'npl-1.1': {'id': 'NPL-1.1', 'deprecated': False},
'nposl-3.0': {'id': 'NPOSL-3.0', 'deprecated': False},
'nrl': {'id': 'NRL', 'deprecated': False},
'ntia-pd': {'id': 'NTIA-PD', 'deprecated': False},
'ntp': {'id': 'NTP', 'deprecated': False},
'ntp-0': {'id': 'NTP-0', 'deprecated': False},
'nunit': {'id': 'Nunit', 'deprecated': True},
@@ -580,6 +604,7 @@ LICENSES: dict[str, SPDXLicense] = {
'schemereport': {'id': 'SchemeReport', 'deprecated': False},
'sendmail': {'id': 'Sendmail', 'deprecated': False},
'sendmail-8.23': {'id': 'Sendmail-8.23', 'deprecated': False},
'sendmail-open-source-1.1': {'id': 'Sendmail-Open-Source-1.1', 'deprecated': False},
'sgi-b-1.0': {'id': 'SGI-B-1.0', 'deprecated': False},
'sgi-b-1.1': {'id': 'SGI-B-1.1', 'deprecated': False},
'sgi-b-2.0': {'id': 'SGI-B-2.0', 'deprecated': False},
@@ -592,10 +617,12 @@ LICENSES: dict[str, SPDXLicense] = {
'sissl-1.2': {'id': 'SISSL-1.2', 'deprecated': False},
'sl': {'id': 'SL', 'deprecated': False},
'sleepycat': {'id': 'Sleepycat', 'deprecated': False},
'smail-gpl': {'id': 'SMAIL-GPL', 'deprecated': False},
'smlnj': {'id': 'SMLNJ', 'deprecated': False},
'smppl': {'id': 'SMPPL', 'deprecated': False},
'snia': {'id': 'SNIA', 'deprecated': False},
'snprintf': {'id': 'snprintf', 'deprecated': False},
'sofa': {'id': 'SOFA', 'deprecated': False},
'softsurfer': {'id': 'softSurfer', 'deprecated': False},
'soundex': {'id': 'Soundex', 'deprecated': False},
'spencer-86': {'id': 'Spencer-86', 'deprecated': False},
@@ -609,6 +636,7 @@ LICENSES: dict[str, SPDXLicense] = {
'sspl-1.0': {'id': 'SSPL-1.0', 'deprecated': False},
'standardml-nj': {'id': 'StandardML-NJ', 'deprecated': True},
'sugarcrm-1.1.3': {'id': 'SugarCRM-1.1.3', 'deprecated': False},
'sul-1.0': {'id': 'SUL-1.0', 'deprecated': False},
'sun-ppp': {'id': 'Sun-PPP', 'deprecated': False},
'sun-ppp-2000': {'id': 'Sun-PPP-2000', 'deprecated': False},
'sunpro': {'id': 'SunPro', 'deprecated': False},
@@ -620,12 +648,14 @@ LICENSES: dict[str, SPDXLicense] = {
'tcp-wrappers': {'id': 'TCP-wrappers', 'deprecated': False},
'termreadkey': {'id': 'TermReadKey', 'deprecated': False},
'tgppl-1.0': {'id': 'TGPPL-1.0', 'deprecated': False},
'thirdeye': {'id': 'ThirdEye', 'deprecated': False},
'threeparttable': {'id': 'threeparttable', 'deprecated': False},
'tmate': {'id': 'TMate', 'deprecated': False},
'torque-1.1': {'id': 'TORQUE-1.1', 'deprecated': False},
'tosl': {'id': 'TOSL', 'deprecated': False},
'tpdl': {'id': 'TPDL', 'deprecated': False},
'tpl-1.0': {'id': 'TPL-1.0', 'deprecated': False},
'trustedqsl': {'id': 'TrustedQSL', 'deprecated': False},
'ttwl': {'id': 'TTWL', 'deprecated': False},
'ttyp0': {'id': 'TTYP0', 'deprecated': False},
'tu-berlin-1.0': {'id': 'TU-Berlin-1.0', 'deprecated': False},
@@ -641,6 +671,8 @@ LICENSES: dict[str, SPDXLicense] = {
'unicode-tou': {'id': 'Unicode-TOU', 'deprecated': False},
'unixcrypt': {'id': 'UnixCrypt', 'deprecated': False},
'unlicense': {'id': 'Unlicense', 'deprecated': False},
'unlicense-libtelnet': {'id': 'Unlicense-libtelnet', 'deprecated': False},
'unlicense-libwhirlpool': {'id': 'Unlicense-libwhirlpool', 'deprecated': False},
'upl-1.0': {'id': 'UPL-1.0', 'deprecated': False},
'urt-rle': {'id': 'URT-RLE', 'deprecated': False},
'vim': {'id': 'Vim', 'deprecated': False},
@@ -654,6 +686,7 @@ LICENSES: dict[str, SPDXLicense] = {
'widget-workshop': {'id': 'Widget-Workshop', 'deprecated': False},
'wsuipa': {'id': 'Wsuipa', 'deprecated': False},
'wtfpl': {'id': 'WTFPL', 'deprecated': False},
'wwl': {'id': 'wwl', 'deprecated': False},
'wxwindows': {'id': 'wxWindows', 'deprecated': True},
'x11': {'id': 'X11', 'deprecated': False},
'x11-distribute-modifications-variant': {'id': 'X11-distribute-modifications-variant', 'deprecated': False},
@@ -695,9 +728,11 @@ EXCEPTIONS: dict[str, SPDXException] = {
'bison-exception-1.24': {'id': 'Bison-exception-1.24', 'deprecated': False},
'bison-exception-2.2': {'id': 'Bison-exception-2.2', 'deprecated': False},
'bootloader-exception': {'id': 'Bootloader-exception', 'deprecated': False},
'cgal-linking-exception': {'id': 'CGAL-linking-exception', 'deprecated': False},
'classpath-exception-2.0': {'id': 'Classpath-exception-2.0', 'deprecated': False},
'clisp-exception-2.0': {'id': 'CLISP-exception-2.0', 'deprecated': False},
'cryptsetup-openssl-exception': {'id': 'cryptsetup-OpenSSL-exception', 'deprecated': False},
'digia-qt-lgpl-exception-1.1': {'id': 'Digia-Qt-LGPL-exception-1.1', 'deprecated': False},
'digirule-foss-exception': {'id': 'DigiRule-FOSS-exception', 'deprecated': False},
'ecos-exception-2.0': {'id': 'eCos-exception-2.0', 'deprecated': False},
'erlang-otp-linking-exception': {'id': 'erlang-otp-linking-exception', 'deprecated': False},
@@ -714,13 +749,16 @@ EXCEPTIONS: dict[str, SPDXException] = {
'gnome-examples-exception': {'id': 'GNOME-examples-exception', 'deprecated': False},
'gnu-compiler-exception': {'id': 'GNU-compiler-exception', 'deprecated': False},
'gnu-javamail-exception': {'id': 'gnu-javamail-exception', 'deprecated': False},
'gpl-3.0-389-ds-base-exception': {'id': 'GPL-3.0-389-ds-base-exception', 'deprecated': False},
'gpl-3.0-interface-exception': {'id': 'GPL-3.0-interface-exception', 'deprecated': False},
'gpl-3.0-linking-exception': {'id': 'GPL-3.0-linking-exception', 'deprecated': False},
'gpl-3.0-linking-source-exception': {'id': 'GPL-3.0-linking-source-exception', 'deprecated': False},
'gpl-cc-1.0': {'id': 'GPL-CC-1.0', 'deprecated': False},
'gstreamer-exception-2005': {'id': 'GStreamer-exception-2005', 'deprecated': False},
'gstreamer-exception-2008': {'id': 'GStreamer-exception-2008', 'deprecated': False},
'harbour-exception': {'id': 'harbour-exception', 'deprecated': False},
'i2p-gpl-java-exception': {'id': 'i2p-gpl-java-exception', 'deprecated': False},
'independent-modules-exception': {'id': 'Independent-modules-exception', 'deprecated': False},
'kicad-libraries-exception': {'id': 'KiCad-libraries-exception', 'deprecated': False},
'lgpl-3.0-linking-exception': {'id': 'LGPL-3.0-linking-exception', 'deprecated': False},
'libpri-openh323-exception': {'id': 'libpri-OpenH323-exception', 'deprecated': False},
@@ -730,12 +768,14 @@ EXCEPTIONS: dict[str, SPDXException] = {
'llvm-exception': {'id': 'LLVM-exception', 'deprecated': False},
'lzma-exception': {'id': 'LZMA-exception', 'deprecated': False},
'mif-exception': {'id': 'mif-exception', 'deprecated': False},
'mxml-exception': {'id': 'mxml-exception', 'deprecated': False},
'nokia-qt-exception-1.1': {'id': 'Nokia-Qt-exception-1.1', 'deprecated': True},
'ocaml-lgpl-linking-exception': {'id': 'OCaml-LGPL-linking-exception', 'deprecated': False},
'occt-exception-1.0': {'id': 'OCCT-exception-1.0', 'deprecated': False},
'openjdk-assembly-exception-1.0': {'id': 'OpenJDK-assembly-exception-1.0', 'deprecated': False},
'openvpn-openssl-exception': {'id': 'openvpn-openssl-exception', 'deprecated': False},
'pcre2-exception': {'id': 'PCRE2-exception', 'deprecated': False},
'polyparse-exception': {'id': 'polyparse-exception', 'deprecated': False},
'ps-or-pdf-font-exception-20170817': {'id': 'PS-or-PDF-font-exception-20170817', 'deprecated': False},
'qpl-1.0-inria-2004-exception': {'id': 'QPL-1.0-INRIA-2004-exception', 'deprecated': False},
'qt-gpl-exception-1.0': {'id': 'Qt-GPL-exception-1.0', 'deprecated': False},
+185 -55
View File
@@ -8,7 +8,7 @@ import operator
import os
import platform
import sys
from typing import AbstractSet, Any, Callable, Literal, TypedDict, Union, cast
from typing import AbstractSet, Callable, Literal, Mapping, TypedDict, Union, cast
from ._parser import MarkerAtom, MarkerList, Op, Value, Variable
from ._parser import parse_marker as _parse_marker
@@ -17,6 +17,7 @@ from .specifiers import InvalidSpecifier, Specifier
from .utils import canonicalize_name
__all__ = [
"Environment",
"EvaluateContext",
"InvalidMarker",
"Marker",
@@ -25,31 +26,58 @@ __all__ = [
"default_environment",
]
def __dir__() -> list[str]:
return __all__
Operator = Callable[[str, Union[str, AbstractSet[str]]], bool]
EvaluateContext = Literal["metadata", "lock_file", "requirement"]
"""A ``typing.Literal`` enumerating valid marker evaluation contexts.
Valid values for the ``context`` passed to :meth:`Marker.evaluate` are:
* ``"metadata"`` (for core metadata; default)
* ``"lock_file"`` (for lock files)
* ``"requirement"`` (i.e. all other situations)
"""
MARKERS_ALLOWING_SET = {"extras", "dependency_groups"}
MARKERS_REQUIRING_VERSION = {
"implementation_version",
"platform_release",
"python_full_version",
"python_version",
}
class InvalidMarker(ValueError):
"""
An invalid marker was found, users should refer to PEP 508.
"""Raised when attempting to create a :class:`Marker` from invalid input.
This error indicates that the given marker string does not conform to the
:ref:`specification of dependency specifiers <pypug:dependency-specifiers>`.
"""
class UndefinedComparison(ValueError):
"""
An invalid operation was attempted on a value that doesn't support it.
"""Raised when evaluating an unsupported marker comparison.
This can happen when marker values are compared as versions but do not
conform to the :ref:`specification of version specifiers
<pypug:version-specifiers>`.
"""
class UndefinedEnvironmentName(ValueError):
"""
A name was attempted to be used that does not exist inside of the
environment.
"""
"""Raised when evaluating a marker that references a missing environment key."""
class Environment(TypedDict):
"""
A dictionary that represents a Python environment as captured by
:func:`default_environment`. All fields are required.
"""
implementation_name: str
"""The implementation's identifier, e.g. ``'cpython'``."""
@@ -121,20 +149,28 @@ class Environment(TypedDict):
"""
def _normalize_extra_values(results: Any) -> Any:
def _normalize_extras(
result: MarkerList | MarkerAtom | str,
) -> MarkerList | MarkerAtom | str:
if not isinstance(result, tuple):
return result
lhs, op, rhs = result
if isinstance(lhs, Variable) and lhs.value == "extra":
normalized_extra = canonicalize_name(rhs.value)
rhs = Value(normalized_extra)
elif isinstance(rhs, Variable) and rhs.value == "extra":
normalized_extra = canonicalize_name(lhs.value)
lhs = Value(normalized_extra)
return lhs, op, rhs
def _normalize_extra_values(results: MarkerList) -> MarkerList:
"""
Normalize extra values.
"""
if isinstance(results[0], tuple):
lhs, op, rhs = results[0]
if isinstance(lhs, Variable) and lhs.value == "extra":
normalized_extra = canonicalize_name(rhs.value)
rhs = Value(normalized_extra)
elif isinstance(rhs, Variable) and rhs.value == "extra":
normalized_extra = canonicalize_name(lhs.value)
lhs = Value(normalized_extra)
results[0] = lhs, op, rhs
return results
return [_normalize_extras(r) for r in results]
def _format_marker(
@@ -168,25 +204,26 @@ def _format_marker(
_operators: dict[str, Operator] = {
"in": lambda lhs, rhs: lhs in rhs,
"not in": lambda lhs, rhs: lhs not in rhs,
"<": operator.lt,
"<=": operator.le,
"<": lambda _lhs, _rhs: False,
"<=": operator.eq,
"==": operator.eq,
"!=": operator.ne,
">=": operator.ge,
">": operator.gt,
">=": operator.eq,
">": lambda _lhs, _rhs: False,
}
def _eval_op(lhs: str, op: Op, rhs: str | AbstractSet[str]) -> bool:
if isinstance(rhs, str):
def _eval_op(lhs: str, op: Op, rhs: str | AbstractSet[str], *, key: str) -> bool:
op_str = op.serialize()
if key in MARKERS_REQUIRING_VERSION:
try:
spec = Specifier("".join([op.serialize(), rhs]))
spec = Specifier(f"{op_str}{rhs}")
except InvalidSpecifier:
pass
else:
return spec.contains(lhs, prereleases=True)
oper: Operator | None = _operators.get(op.serialize())
oper: Operator | None = _operators.get(op_str)
if oper is None:
raise UndefinedComparison(f"Undefined {op!r} on {lhs!r} and {rhs!r}.")
@@ -196,13 +233,14 @@ def _eval_op(lhs: str, op: Op, rhs: str | AbstractSet[str]) -> bool:
def _normalize(
lhs: str, rhs: str | AbstractSet[str], key: str
) -> tuple[str, str | AbstractSet[str]]:
# PEP 685 Comparison of extra names for optional distribution dependencies
# PEP 685 - Comparison of extra names for optional distribution dependencies
# https://peps.python.org/pep-0685/
# > When comparing extra names, tools MUST normalize the names being
# > compared using the semantics outlined in PEP 503 for names
if key == "extra":
assert isinstance(rhs, str), "extra value must be a string"
return (canonicalize_name(lhs), canonicalize_name(rhs))
# Both sides are normalized at this point already
return (lhs, rhs)
if key in MARKERS_ALLOWING_SET:
if isinstance(rhs, str): # pragma: no cover
return (canonicalize_name(lhs), canonicalize_name(rhs))
@@ -219,8 +257,6 @@ def _evaluate_markers(
groups: list[list[bool]] = [[]]
for marker in markers:
assert isinstance(marker, (list, tuple, str))
if isinstance(marker, list):
groups[-1].append(_evaluate_markers(marker, environment))
elif isinstance(marker, tuple):
@@ -234,18 +270,21 @@ def _evaluate_markers(
lhs_value = lhs.value
environment_key = rhs.value
rhs_value = environment[environment_key]
assert isinstance(lhs_value, str), "lhs must be a string"
lhs_value, rhs_value = _normalize(lhs_value, rhs_value, key=environment_key)
groups[-1].append(_eval_op(lhs_value, op, rhs_value))
else:
assert marker in ["and", "or"]
if marker == "or":
groups.append([])
groups[-1].append(_eval_op(lhs_value, op, rhs_value, key=environment_key))
elif marker == "or":
groups.append([])
elif marker == "and":
pass
else: # pragma: nocover
raise TypeError(f"Unexpected marker {marker!r}")
return any(all(item) for item in groups)
def format_full_version(info: sys._version_info) -> str:
def _format_full_version(info: sys._version_info) -> str:
version = f"{info.major}.{info.minor}.{info.micro}"
kind = info.releaselevel
if kind != "final":
@@ -254,7 +293,11 @@ def format_full_version(info: sys._version_info) -> str:
def default_environment() -> Environment:
iver = format_full_version(sys.implementation.version)
"""Return the default marker environment for the current Python process.
This is the base environment used by :meth:`Marker.evaluate`.
"""
iver = _format_full_version(sys.implementation.version)
implementation_name = sys.implementation.name
return {
"implementation_name": implementation_name,
@@ -272,10 +315,36 @@ def default_environment() -> Environment:
class Marker:
"""Represents a parsed dependency marker expression.
Marker expressions are parsed according to the
:ref:`specification of dependency specifiers <pypug:dependency-specifiers>`.
:param marker: The string representation of a marker expression.
:raises InvalidMarker: If ``marker`` cannot be parsed.
Instances are safe to serialize with :mod:`pickle`. They use a stable
format so the same pickle can be loaded in future packaging releases.
.. versionchanged:: 26.2
Added a stable pickle format. Pickles created with packaging 26.2+ can
be unpickled with future releases. Backward compatibility with pickles
from packaging < 26.2 is supported but may be removed in a future
release.
"""
__slots__ = ("_markers",)
def __init__(self, marker: str) -> None:
# Note: We create a Marker object without calling this constructor in
# packaging.requirements.Requirement. If any additional logic is
# added here, make sure to mirror/adapt Requirement.
# If this fails and throws an error, the repr still expects _markers to
# be defined.
self._markers: MarkerList = []
try:
self._markers = _normalize_extra_values(_parse_marker(marker))
# The attribute `_markers` can be described in terms of a recursive type:
@@ -297,36 +366,94 @@ class Marker:
except ParserSyntaxError as e:
raise InvalidMarker(str(e)) from e
@classmethod
def _from_markers(cls, markers: MarkerList) -> Marker:
"""Create a Marker instance from a pre-parsed marker tree.
This avoids re-parsing serialised marker strings when combining markers.
"""
new = cls.__new__(cls)
new._markers = markers
return new
def __str__(self) -> str:
return _format_marker(self._markers)
def __repr__(self) -> str:
return f"<Marker('{self}')>"
return f"<{self.__class__.__name__}({str(self)!r})>"
def __hash__(self) -> int:
return hash((self.__class__.__name__, str(self)))
return hash(str(self))
def __eq__(self, other: Any) -> bool:
def __eq__(self, other: object) -> bool:
if not isinstance(other, Marker):
return NotImplemented
return str(self) == str(other)
def __getstate__(self) -> str:
# Return the marker expression string for compactness and stability.
# Internal Node objects are excluded; the string is re-parsed on load.
return str(self)
def __setstate__(self, state: object) -> None:
if isinstance(state, str):
# New format (26.2+): just the marker expression string.
try:
self._markers = _normalize_extra_values(_parse_marker(state))
except ParserSyntaxError as exc:
raise TypeError(f"Cannot restore Marker from {state!r}") from exc
return
if isinstance(state, dict) and "_markers" in state:
# Old format (packaging <= 26.1, no __slots__): plain __dict__.
markers = state["_markers"]
if isinstance(markers, list):
self._markers = markers
return
if isinstance(state, tuple) and len(state) == 2:
# Old format (packaging <= 26.1, __slots__): (None, {slot: value}).
_, slot_dict = state
if isinstance(slot_dict, dict) and "_markers" in slot_dict:
markers = slot_dict["_markers"]
if isinstance(markers, list):
self._markers = markers
return
raise TypeError(f"Cannot restore Marker from {state!r}")
def __and__(self, other: Marker) -> Marker:
if not isinstance(other, Marker):
return NotImplemented
return self._from_markers([self._markers, "and", other._markers])
def __or__(self, other: Marker) -> Marker:
if not isinstance(other, Marker):
return NotImplemented
return self._from_markers([self._markers, "or", other._markers])
def evaluate(
self,
environment: dict[str, str] | None = None,
environment: Mapping[str, str | AbstractSet[str]] | None = None,
context: EvaluateContext = "metadata",
) -> bool:
"""Evaluate a marker.
Return the boolean from evaluating the given marker against the
environment. environment is an optional argument to override all or
part of the determined environment. The *context* parameter specifies what
context the markers are being evaluated for, which influences what markers
are considered valid. Acceptable values are "metadata" (for core metadata;
default), "lock_file", and "requirement" (i.e. all other situations).
Return the boolean from evaluating this marker against the environment.
The environment is determined from the current Python process unless
passed in explicitly.
:param environment: Mapping containing keys and values to override the
detected environment.
:param EvaluateContext context: The context in which the marker is
evaluated, which influences what marker names are considered valid.
Accepted values are ``"metadata"`` (for core metadata; default),
``"lock_file"``, and ``"requirement"`` (i.e. all other situations).
:raises UndefinedComparison: If the marker uses a comparison on values
that are not valid versions per the :ref:`specification of version
specifiers <pypug:version-specifiers>`.
:raises UndefinedEnvironmentName: If the marker references a value that
is missing from the evaluation environment.
:returns: ``True`` if the marker matches, otherwise ``False``.
The environment is determined from the current Python process.
"""
current_environment = cast(
"dict[str, str | AbstractSet[str]]", default_environment()
@@ -337,12 +464,15 @@ class Marker:
)
elif context == "metadata":
current_environment["extra"] = ""
if environment is not None:
current_environment.update(environment)
# The API used to allow setting extra to None. We need to handle this
# case for backwards compatibility.
if "extra" in current_environment and current_environment["extra"] is None:
current_environment["extra"] = ""
if "extra" in current_environment:
# The API used to allow setting extra to None. We need to handle
# this case for backwards compatibility. Also skip running
# normalize name if extra is empty.
extra = cast("str | None", current_environment["extra"])
current_environment["extra"] = canonicalize_name(extra) if extra else ""
return _evaluate_markers(
self._markers, _repair_python_full_version(current_environment)
@@ -356,7 +486,7 @@ def _repair_python_full_version(
Work around platform.python_version() returning something that is not PEP 440
compliant for non-tagged Python builds.
"""
python_full_version = cast(str, env["python_full_version"])
python_full_version = cast("str", env["python_full_version"])
if python_full_version.endswith("+"):
env["python_full_version"] = f"{python_full_version}local"
return env
+171 -69
View File
@@ -1,12 +1,11 @@
from __future__ import annotations
import email.feedparser
import email.header
import email.message
import email.parser
import email.policy
import keyword
import pathlib
import sys
import typing
from typing import (
Any,
@@ -19,31 +18,27 @@ from typing import (
from . import licenses, requirements, specifiers, utils
from . import version as version_module
from .licenses import NormalizedLicenseExpression
from .errors import ExceptionGroup, _ErrorCollector
if typing.TYPE_CHECKING:
from .licenses import NormalizedLicenseExpression
T = typing.TypeVar("T")
if sys.version_info >= (3, 11): # pragma: no cover
ExceptionGroup = ExceptionGroup
else: # pragma: no cover
__all__ = [
"ExceptionGroup", # Keep this for a bit (makes mypy happy w/ 26.0 compat)
"InvalidMetadata",
"Metadata",
"RFC822Message",
"RFC822Policy",
"RawMetadata",
"parse_email",
]
class ExceptionGroup(Exception):
"""A minimal implementation of :external:exc:`ExceptionGroup` from Python 3.11.
If :external:exc:`ExceptionGroup` is already defined by Python itself,
that version is used instead.
"""
message: str
exceptions: list[Exception]
def __init__(self, message: str, exceptions: list[Exception]) -> None:
self.message = message
self.exceptions = exceptions
def __repr__(self) -> str:
return f"{self.__class__.__name__}({self.message!r}, {self.exceptions!r})"
def __dir__() -> list[str]:
return __all__
class InvalidMetadata(ValueError):
@@ -73,8 +68,8 @@ class RawMetadata(TypedDict, total=False):
Core metadata fields that can be specified multiple times are stored as a
list or dict depending on which is appropriate for the field. Any fields
which hold multiple values in a single field are stored as a list.
which hold multiple values in a single field are stored as a list. All fields
are considered optional.
"""
# Metadata 1.0 - PEP 241
@@ -126,13 +121,19 @@ class RawMetadata(TypedDict, total=False):
# Metadata 2.3 - PEP 685
# No new fields were added in PEP 685, just some edge case were
# tightened up to provide better interoptability.
# tightened up to provide better interoperability.
# Metadata 2.4 - PEP 639
license_expression: str
license_files: list[str]
# Metadata 2.5 - PEP 794
import_names: list[str]
import_namespaces: list[str]
# 'keywords' is special as it's a string in the core metadata spec, but we
# represent it as a list.
_STRING_FIELDS = {
"author",
"author_email",
@@ -165,6 +166,8 @@ _LIST_FIELDS = {
"requires_dist",
"requires_external",
"supported_platforms",
"import_names",
"import_namespaces",
}
_DICT_FIELDS = {
@@ -193,24 +196,23 @@ def _parse_project_urls(data: list[str]) -> dict[str, str]:
# be the missing value, then they'd have multiple '' values that
# overwrite each other in a accumulating dict.
#
# The other potentional issue is that it's possible to have the
# The other potential issue is that it's possible to have the
# same label multiple times in the metadata, with no solid "right"
# answer with what to do in that case. As such, we'll do the only
# thing we can, which is treat the field as unparseable and add it
# thing we can, which is treat the field as unparsable and add it
# to our list of unparsed fields.
parts = [p.strip() for p in pair.split(",", 1)]
parts.extend([""] * (max(0, 2 - len(parts)))) # Ensure 2 items
#
# TODO: The spec doesn't say anything about if the keys should be
# considered case sensitive or not... logically they should
# be case-preserving and case-insensitive, but doing that
# would open up more cases where we might have duplicate
# entries.
label, url = parts
label, _, url = (s.strip() for s in pair.partition(","))
if label in urls:
# The label already exists in our set of urls, so this field
# is unparseable, and we can just add the whole thing to our
# unparseable data and stop processing it.
# is unparsable, and we can just add the whole thing to our
# unparsable data and stop processing it.
raise KeyError("duplicate labels in project urls")
urls[label] = url
@@ -257,6 +259,8 @@ _EMAIL_TO_RAW_MAPPING = {
"download-url": "download_url",
"dynamic": "dynamic",
"home-page": "home_page",
"import-name": "import_names",
"import-namespace": "import_namespaces",
"keywords": "keywords",
"license": "license",
"license-expression": "license_expression",
@@ -283,6 +287,45 @@ _EMAIL_TO_RAW_MAPPING = {
_RAW_TO_EMAIL_MAPPING = {raw: email for email, raw in _EMAIL_TO_RAW_MAPPING.items()}
# This class is for writing RFC822 messages
class RFC822Policy(email.policy.EmailPolicy):
"""
This is :class:`email.policy.EmailPolicy`, but with a simple ``header_store_parse``
implementation that handles multi-line values, and some nice defaults.
"""
utf8 = True
mangle_from_ = False
max_line_length = 0
def header_store_parse(self, name: str, value: str) -> tuple[str, str]:
size = len(name) + 2
value = value.replace("\n", "\n" + " " * size)
return (name, value)
# This class is for writing RFC822 messages
class RFC822Message(email.message.EmailMessage):
"""
This is :class:`email.message.EmailMessage` with two small changes: it defaults to
our `RFC822Policy`, and it correctly writes unicode when being called
with `bytes()`.
"""
def __init__(self) -> None:
super().__init__(policy=RFC822Policy())
def as_bytes(
self, unixfrom: bool = False, policy: email.policy.Policy | None = None
) -> bytes:
"""
Return the bytes representation of the message.
This handles unicode encoding.
"""
return self.as_string(unixfrom, policy=policy).encode("utf-8")
def parse_email(data: bytes | str) -> tuple[RawMetadata, dict[str, list[str]]]:
"""Parse a distribution's metadata stored as email headers (e.g. from ``METADATA``).
@@ -310,10 +353,10 @@ def parse_email(data: bytes | str) -> tuple[RawMetadata, dict[str, list[str]]]:
# We have to wrap parsed.keys() in a set, because in the case of multiple
# values for a key (a list), the key will appear multiple times in the
# list of keys, but we're avoiding that by using get_all().
for name in frozenset(parsed.keys()):
for name_with_case in frozenset(parsed.keys()):
# Header names in RFC are case insensitive, so we'll normalize to all
# lower case to make comparisons easier.
name = name.lower()
name = name_with_case.lower()
# We use get_all() here, even for fields that aren't multiple use,
# because otherwise someone could have e.g. two Name fields, and we
@@ -349,16 +392,16 @@ def parse_email(data: bytes | str) -> tuple[RawMetadata, dict[str, list[str]]]:
# can be independently encoded, so we'll need to check each
# of them.
chunks: list[tuple[bytes, str | None]] = []
for bin, encoding in email.header.decode_header(h):
for binary, _encoding in email.header.decode_header(h):
try:
bin.decode("utf8", "strict")
binary.decode("utf8", "strict")
except UnicodeDecodeError:
# Enable mojibake.
encoding = "latin1"
valid_encoding = False
else:
encoding = "utf8"
chunks.append((bin, encoding))
chunks.append((binary, encoding))
# Turn our chunks back into a Header object, then let that
# Header object do the right thing to turn them into a
@@ -397,6 +440,11 @@ def parse_email(data: bytes | str) -> tuple[RawMetadata, dict[str, list[str]]]:
# of unparsed stuff.
if raw_name in _STRING_FIELDS and len(value) == 1:
raw[raw_name] = value[0]
# If this is import_names, we need to special case the empty field
# case, which converts to an empty list instead of None. We can't let
# the empty case slip through, as it will fail validation.
elif raw_name == "import_names" and value == [""]:
raw[raw_name] = []
# If this is one of our list of string fields, then we can just assign
# the value, since email *only* has strings, and our get_all() call
# above ensures that this is a list.
@@ -424,7 +472,7 @@ def parse_email(data: bytes | str) -> tuple[RawMetadata, dict[str, list[str]]]:
except KeyError:
unparsed[name] = value
# Nothing that we've done has managed to parse this, so it'll just
# throw it in our unparseable data and move on.
# throw it in our unparsable data and move on.
else:
unparsed[name] = value
@@ -441,9 +489,9 @@ def parse_email(data: bytes | str) -> tuple[RawMetadata, dict[str, list[str]]]:
else:
if payload:
# Check to see if we've already got a description, if so then both
# it, and this body move to unparseable.
# it, and this body move to unparsable.
if "description" in raw:
description_header = cast(str, raw.pop("description"))
description_header = cast("str", raw.pop("description"))
unparsed.setdefault("description", []).extend(
[description_header, payload]
)
@@ -456,15 +504,15 @@ def parse_email(data: bytes | str) -> tuple[RawMetadata, dict[str, list[str]]]:
# literal key names, but we're computing our key names on purpose, but the
# way this function is implemented, our `TypedDict` can only have valid key
# names.
return cast(RawMetadata, raw), unparsed
return cast("RawMetadata", raw), unparsed
_NOT_FOUND = object()
# Keep the two values in sync.
_VALID_METADATA_VERSIONS = ["1.0", "1.1", "1.2", "2.1", "2.2", "2.3", "2.4"]
_MetadataVersion = Literal["1.0", "1.1", "1.2", "2.1", "2.2", "2.3", "2.4"]
_VALID_METADATA_VERSIONS = ["1.0", "1.1", "1.2", "2.1", "2.2", "2.3", "2.4", "2.5"]
_MetadataVersion = Literal["1.0", "1.1", "1.2", "2.1", "2.2", "2.3", "2.4", "2.5"]
_REQUIRED_ATTRS = frozenset(["metadata_version", "name", "version"])
@@ -519,7 +567,7 @@ class _Validator(Generic[T]):
except KeyError:
pass
return cast(T, value)
return cast("T", value)
def _invalid_metadata(
self, msg: str, cause: Exception | None = None
@@ -534,7 +582,7 @@ class _Validator(Generic[T]):
# Implicitly makes Metadata-Version required.
if value not in _VALID_METADATA_VERSIONS:
raise self._invalid_metadata(f"{value!r} is not a valid metadata version")
return cast(_MetadataVersion, value)
return cast("_MetadataVersion", value)
def _process_name(self, value: str) -> str:
if not value:
@@ -585,7 +633,7 @@ class _Validator(Generic[T]):
charset = parameters.get("charset", "UTF-8")
if charset != "UTF-8":
raise self._invalid_metadata(
f"{{field}} can only specify the UTF-8 charset, not {list(charset)}"
f"{{field}} can only specify the UTF-8 charset, not {charset!r}"
)
markdown_variants = {"GFM", "CommonMark"}
@@ -647,9 +695,7 @@ class _Validator(Generic[T]):
else:
return reqs
def _process_license_expression(
self, value: str
) -> NormalizedLicenseExpression | None:
def _process_license_expression(self, value: str) -> NormalizedLicenseExpression:
try:
return licenses.canonicalize_license_expression(value)
except ValueError as exc:
@@ -683,6 +729,30 @@ class _Validator(Generic[T]):
paths.append(path)
return paths
def _process_import_names(self, value: list[str]) -> list[str]:
for import_name in value:
name, semicolon, private = import_name.partition(";")
name = name.rstrip()
for identifier in name.split("."):
if not identifier.isidentifier():
raise self._invalid_metadata(
f"{name!r} is invalid for {{field}}; "
f"{identifier!r} is not a valid identifier"
)
elif keyword.iskeyword(identifier):
raise self._invalid_metadata(
f"{name!r} is invalid for {{field}}; "
f"{identifier!r} is a keyword"
)
if semicolon and private.lstrip() != "private":
raise self._invalid_metadata(
f"{import_name!r} is invalid for {{field}}; "
"the only valid option is 'private'"
)
return value
_process_import_namespaces = _process_import_names
class Metadata:
"""Representation of distribution metadata.
@@ -706,13 +776,11 @@ class Metadata:
ins._raw = data.copy() # Mutations occur due to caching enriched values.
if validate:
exceptions: list[Exception] = []
try:
collector = _ErrorCollector()
metadata_version = None
with collector.collect(InvalidMetadata):
metadata_version = ins.metadata_version
metadata_age = _VALID_METADATA_VERSIONS.index(metadata_version)
except InvalidMetadata as metadata_version_exc:
exceptions.append(metadata_version_exc)
metadata_version = None
# Make sure to check for the fields that are present, the required
# fields (so their absence can be reported).
@@ -729,7 +797,7 @@ class Metadata:
field_metadata_version = cls.__dict__[key].added
except KeyError:
exc = InvalidMetadata(key, f"unrecognized field: {key!r}")
exceptions.append(exc)
collector.error(exc)
continue
field_age = _VALID_METADATA_VERSIONS.index(
field_metadata_version
@@ -741,14 +809,13 @@ class Metadata:
f"{field} introduced in metadata version "
f"{field_metadata_version}, not {metadata_version}",
)
exceptions.append(exc)
collector.error(exc)
continue
getattr(ins, key)
except InvalidMetadata as exc:
exceptions.append(exc)
collector.error(exc)
if exceptions:
raise ExceptionGroup("invalid metadata", exceptions)
collector.finalize("invalid metadata")
return ins
@@ -762,16 +829,13 @@ class Metadata:
raw, unparsed = parse_email(data)
if validate:
exceptions: list[Exception] = []
for unparsed_key in unparsed:
if unparsed_key in _EMAIL_TO_RAW_MAPPING:
message = f"{unparsed_key!r} has invalid data"
else:
message = f"unrecognized field: {unparsed_key!r}"
exceptions.append(InvalidMetadata(unparsed_key, message))
if exceptions:
raise ExceptionGroup("unparsed", exceptions)
with _ErrorCollector().on_exit("unparsed") as collector:
for unparsed_key in unparsed:
if unparsed_key in _EMAIL_TO_RAW_MAPPING:
message = f"{unparsed_key!r} has invalid data"
else:
message = f"unrecognized field: {unparsed_key!r}"
collector.error(InvalidMetadata(unparsed_key, message))
try:
return cls.from_raw(raw, validate=validate)
@@ -854,9 +918,47 @@ class Metadata:
""":external:ref:`core-metadata-provides-dist`"""
obsoletes_dist: _Validator[list[str] | None] = _Validator(added="1.2")
""":external:ref:`core-metadata-obsoletes-dist`"""
import_names: _Validator[list[str] | None] = _Validator(added="2.5")
""":external:ref:`core-metadata-import-name`"""
import_namespaces: _Validator[list[str] | None] = _Validator(added="2.5")
""":external:ref:`core-metadata-import-namespace`"""
requires: _Validator[list[str] | None] = _Validator(added="1.1")
"""``Requires`` (deprecated)"""
provides: _Validator[list[str] | None] = _Validator(added="1.1")
"""``Provides`` (deprecated)"""
obsoletes: _Validator[list[str] | None] = _Validator(added="1.1")
"""``Obsoletes`` (deprecated)"""
def as_rfc822(self) -> RFC822Message:
"""
Return an RFC822 message with the metadata.
"""
message = RFC822Message()
self._write_metadata(message)
return message
def _write_metadata(self, message: RFC822Message) -> None:
"""
Return an RFC822 message with the metadata.
"""
for name, validator in self.__class__.__dict__.items():
if isinstance(validator, _Validator) and name != "description":
value = getattr(self, name)
email_name = _RAW_TO_EMAIL_MAPPING[name]
if value is not None:
if email_name == "project-url":
for label, url in value.items():
message[email_name] = f"{label}, {url}"
elif email_name == "keywords":
message[email_name] = ",".join(value)
elif email_name == "import-name" and value == []:
message[email_name] = ""
elif isinstance(value, list):
for item in value:
message[email_name] = str(item)
else:
message[email_name] = str(value)
# The description is a special case because it is in the body of the message.
if self.description is not None:
message.set_payload(self.description)
+905
View File
@@ -0,0 +1,905 @@
from __future__ import annotations
import dataclasses
import logging
import re
from collections.abc import Mapping, Sequence
from dataclasses import dataclass
from datetime import datetime
from typing import (
TYPE_CHECKING,
Any,
Callable,
Protocol,
TypeVar,
cast,
)
from urllib.parse import urlparse
from .markers import Environment, Marker, default_environment
from .specifiers import SpecifierSet
from .tags import create_compatible_tags_selector, sys_tags
from .utils import (
NormalizedName,
is_normalized_name,
parse_sdist_filename,
parse_wheel_filename,
)
from .version import Version
if TYPE_CHECKING: # pragma: no cover
from collections.abc import Collection, Iterator
from pathlib import Path
from typing_extensions import Self
from .tags import Tag
_logger = logging.getLogger(__name__)
__all__ = [
"Package",
"PackageArchive",
"PackageDirectory",
"PackageSdist",
"PackageVcs",
"PackageWheel",
"Pylock",
"PylockUnsupportedVersionError",
"PylockValidationError",
"is_valid_pylock_path",
]
def __dir__() -> list[str]:
return __all__
_T = TypeVar("_T")
_T2 = TypeVar("_T2")
class _FromMappingProtocol(Protocol): # pragma: no cover
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self: ...
_FromMappingProtocolT = TypeVar("_FromMappingProtocolT", bound=_FromMappingProtocol)
_PYLOCK_FILE_NAME_RE = re.compile(r"^pylock\.([^.]+)\.toml$")
def is_valid_pylock_path(path: Path) -> bool:
"""Check if the given path is a valid pylock file path."""
return path.name == "pylock.toml" or bool(_PYLOCK_FILE_NAME_RE.match(path.name))
def _toml_key(key: str) -> str:
return key.replace("_", "-")
def _toml_value(key: str, value: Any) -> Any: # noqa: ANN401
if isinstance(value, (Version, Marker, SpecifierSet)):
return str(value)
if isinstance(value, Sequence) and key == "environments":
return [str(v) for v in value]
return value
def _toml_dict_factory(data: list[tuple[str, Any]]) -> dict[str, Any]:
return {
_toml_key(key): _toml_value(key, value)
for key, value in data
if value is not None
}
def _get(d: Mapping[str, Any], expected_type: type[_T], key: str) -> _T | None:
"""Get a value from the dictionary and verify it's the expected type."""
if (value := d.get(key)) is None:
return None
if not isinstance(value, expected_type):
raise PylockValidationError(
f"Unexpected type {type(value).__name__} "
f"(expected {expected_type.__name__})",
context=key,
)
return value
def _get_required(d: Mapping[str, Any], expected_type: type[_T], key: str) -> _T:
"""Get a required value from the dictionary and verify it's the expected type."""
if (value := _get(d, expected_type, key)) is None:
raise _PylockRequiredKeyError(key)
return value
def _get_sequence(
d: Mapping[str, Any], expected_item_type: type[_T], key: str
) -> Sequence[_T] | None:
"""Get a list value from the dictionary and verify it's the expected items type."""
if (value := _get(d, Sequence, key)) is None: # type: ignore[type-abstract]
return None
if isinstance(value, (str, bytes)):
# special case: str and bytes are Sequences, but we want to reject it
raise PylockValidationError(
f"Unexpected type {type(value).__name__} (expected Sequence)",
context=key,
)
for i, item in enumerate(value):
if not isinstance(item, expected_item_type):
raise PylockValidationError(
f"Unexpected type {type(item).__name__} "
f"(expected {expected_item_type.__name__})",
context=f"{key}[{i}]",
)
return value
def _get_as(
d: Mapping[str, Any],
expected_type: type[_T],
target_type: Callable[[_T], _T2],
key: str,
) -> _T2 | None:
"""Get a value from the dictionary, verify it's the expected type,
and convert to the target type.
This assumes the target_type constructor accepts the value.
"""
if (value := _get(d, expected_type, key)) is None:
return None
try:
return target_type(value)
except Exception as e:
raise PylockValidationError(e, context=key) from e
def _get_required_as(
d: Mapping[str, Any],
expected_type: type[_T],
target_type: Callable[[_T], _T2],
key: str,
) -> _T2:
"""Get a required value from the dict, verify it's the expected type,
and convert to the target type."""
if (value := _get_as(d, expected_type, target_type, key)) is None:
raise _PylockRequiredKeyError(key)
return value
def _get_sequence_as(
d: Mapping[str, Any],
expected_item_type: type[_T],
target_item_type: Callable[[_T], _T2],
key: str,
) -> list[_T2] | None:
"""Get list value from dictionary and verify expected items type."""
if (value := _get_sequence(d, expected_item_type, key)) is None:
return None
result = []
try:
for item in value:
typed_item = target_item_type(item)
result.append(typed_item)
except Exception as e:
raise PylockValidationError(e, context=f"{key}[{len(result)}]") from e
return result
def _get_object(
d: Mapping[str, Any], target_type: type[_FromMappingProtocolT], key: str
) -> _FromMappingProtocolT | None:
"""Get a dictionary value from the dictionary and convert it to a dataclass."""
if (value := _get(d, Mapping, key)) is None: # type: ignore[type-abstract]
return None
try:
return target_type._from_dict(value)
except Exception as e:
raise PylockValidationError(e, context=key) from e
def _get_sequence_of_objects(
d: Mapping[str, Any], target_item_type: type[_FromMappingProtocolT], key: str
) -> list[_FromMappingProtocolT] | None:
"""Get a list value from the dictionary and convert its items to a dataclass."""
if (value := _get_sequence(d, Mapping, key)) is None: # type: ignore[type-abstract]
return None
result: list[_FromMappingProtocolT] = []
try:
for item in value:
typed_item = target_item_type._from_dict(item)
result.append(typed_item)
except Exception as e:
raise PylockValidationError(e, context=f"{key}[{len(result)}]") from e
return result
def _get_required_sequence_of_objects(
d: Mapping[str, Any], target_item_type: type[_FromMappingProtocolT], key: str
) -> Sequence[_FromMappingProtocolT]:
"""Get a required list value from the dictionary and convert its items to a
dataclass."""
if (result := _get_sequence_of_objects(d, target_item_type, key)) is None:
raise _PylockRequiredKeyError(key)
return result
def _validate_normalized_name(name: str) -> NormalizedName:
"""Validate that a string is a NormalizedName."""
if not is_normalized_name(name):
raise PylockValidationError(f"Name {name!r} is not normalized")
return NormalizedName(name)
def _validate_path_url(path: str | None, url: str | None) -> None:
if not path and not url:
raise PylockValidationError("path or url must be provided")
def _path_name(path: str | None) -> str | None:
if not path:
return None
# If the path is relative it MAY use POSIX-style path separators explicitly
# for portability
if "/" in path:
return path.rsplit("/", 1)[-1]
elif "\\" in path:
return path.rsplit("\\", 1)[-1]
else:
return path
def _url_name(url: str | None) -> str | None:
if not url:
return None
url_path = urlparse(url).path
return url_path.rsplit("/", 1)[-1]
def _validate_hashes(hashes: Mapping[str, Any]) -> Mapping[str, Any]:
if not hashes:
raise PylockValidationError("At least one hash must be provided")
if not all(isinstance(hash_val, str) for hash_val in hashes.values()):
raise PylockValidationError("Hash values must be strings")
return hashes
class PylockValidationError(Exception):
"""Raised when when input data is not spec-compliant."""
context: str | None = None
message: str
def __init__(
self,
cause: str | Exception,
*,
context: str | None = None,
) -> None:
if isinstance(cause, PylockValidationError):
if cause.context:
self.context = (
f"{context}.{cause.context}" if context else cause.context
)
else:
self.context = context
self.message = cause.message
else:
self.context = context
self.message = str(cause)
def __str__(self) -> str:
if self.context:
return f"{self.message} in {self.context!r}"
return self.message
class _PylockRequiredKeyError(PylockValidationError):
def __init__(self, key: str) -> None:
super().__init__("Missing required value", context=key)
class PylockUnsupportedVersionError(PylockValidationError):
"""Raised when encountering an unsupported `lock_version`."""
class PylockSelectError(Exception):
"""Base exception for errors raised by :meth:`Pylock.select`."""
@dataclass(frozen=True, init=False)
class PackageVcs:
type: str
url: str | None = None
path: str | None = None
requested_revision: str | None = None
commit_id: str # type: ignore[misc]
subdirectory: str | None = None
def __init__(
self,
*,
type: str,
url: str | None = None,
path: str | None = None,
requested_revision: str | None = None,
commit_id: str,
subdirectory: str | None = None,
) -> None:
# In Python 3.10+ make dataclass kw_only=True and remove __init__
object.__setattr__(self, "type", type)
object.__setattr__(self, "url", url)
object.__setattr__(self, "path", path)
object.__setattr__(self, "requested_revision", requested_revision)
object.__setattr__(self, "commit_id", commit_id)
object.__setattr__(self, "subdirectory", subdirectory)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
package_vcs = cls(
type=_get_required(d, str, "type"),
url=_get(d, str, "url"),
path=_get(d, str, "path"),
requested_revision=_get(d, str, "requested-revision"),
commit_id=_get_required(d, str, "commit-id"),
subdirectory=_get(d, str, "subdirectory"),
)
_validate_path_url(package_vcs.path, package_vcs.url)
return package_vcs
@dataclass(frozen=True, init=False)
class PackageDirectory:
path: str
editable: bool | None = None
subdirectory: str | None = None
def __init__(
self,
*,
path: str,
editable: bool | None = None,
subdirectory: str | None = None,
) -> None:
# In Python 3.10+ make dataclass kw_only=True and remove __init__
object.__setattr__(self, "path", path)
object.__setattr__(self, "editable", editable)
object.__setattr__(self, "subdirectory", subdirectory)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
return cls(
path=_get_required(d, str, "path"),
editable=_get(d, bool, "editable"),
subdirectory=_get(d, str, "subdirectory"),
)
@dataclass(frozen=True, init=False)
class PackageArchive:
url: str | None = None
path: str | None = None
size: int | None = None
upload_time: datetime | None = None
hashes: Mapping[str, str] # type: ignore[misc]
subdirectory: str | None = None
def __init__(
self,
*,
url: str | None = None,
path: str | None = None,
size: int | None = None,
upload_time: datetime | None = None,
hashes: Mapping[str, str],
subdirectory: str | None = None,
) -> None:
# In Python 3.10+ make dataclass kw_only=True and remove __init__
object.__setattr__(self, "url", url)
object.__setattr__(self, "path", path)
object.__setattr__(self, "size", size)
object.__setattr__(self, "upload_time", upload_time)
object.__setattr__(self, "hashes", hashes)
object.__setattr__(self, "subdirectory", subdirectory)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
package_archive = cls(
url=_get(d, str, "url"),
path=_get(d, str, "path"),
size=_get(d, int, "size"),
upload_time=_get(d, datetime, "upload-time"),
hashes=_get_required_as(d, Mapping, _validate_hashes, "hashes"), # type: ignore[type-abstract]
subdirectory=_get(d, str, "subdirectory"),
)
_validate_path_url(package_archive.path, package_archive.url)
return package_archive
@dataclass(frozen=True, init=False)
class PackageSdist:
name: str | None = None
upload_time: datetime | None = None
url: str | None = None
path: str | None = None
size: int | None = None
hashes: Mapping[str, str] # type: ignore[misc]
def __init__(
self,
*,
name: str | None = None,
upload_time: datetime | None = None,
url: str | None = None,
path: str | None = None,
size: int | None = None,
hashes: Mapping[str, str],
) -> None:
# In Python 3.10+ make dataclass kw_only=True and remove __init__
object.__setattr__(self, "name", name)
object.__setattr__(self, "upload_time", upload_time)
object.__setattr__(self, "url", url)
object.__setattr__(self, "path", path)
object.__setattr__(self, "size", size)
object.__setattr__(self, "hashes", hashes)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
package_sdist = cls(
name=_get(d, str, "name"),
upload_time=_get(d, datetime, "upload-time"),
url=_get(d, str, "url"),
path=_get(d, str, "path"),
size=_get(d, int, "size"),
hashes=_get_required_as(d, Mapping, _validate_hashes, "hashes"), # type: ignore[type-abstract]
)
_validate_path_url(package_sdist.path, package_sdist.url)
return package_sdist
@property
def filename(self) -> str:
"""Get the filename of the sdist."""
filename = self.name or _path_name(self.path) or _url_name(self.url)
if not filename:
raise PylockValidationError("Cannot determine sdist filename")
return filename
@dataclass(frozen=True, init=False)
class PackageWheel:
name: str | None = None
upload_time: datetime | None = None
url: str | None = None
path: str | None = None
size: int | None = None
hashes: Mapping[str, str] # type: ignore[misc]
def __init__(
self,
*,
name: str | None = None,
upload_time: datetime | None = None,
url: str | None = None,
path: str | None = None,
size: int | None = None,
hashes: Mapping[str, str],
) -> None:
# In Python 3.10+ make dataclass kw_only=True and remove __init__
object.__setattr__(self, "name", name)
object.__setattr__(self, "upload_time", upload_time)
object.__setattr__(self, "url", url)
object.__setattr__(self, "path", path)
object.__setattr__(self, "size", size)
object.__setattr__(self, "hashes", hashes)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
package_wheel = cls(
name=_get(d, str, "name"),
upload_time=_get(d, datetime, "upload-time"),
url=_get(d, str, "url"),
path=_get(d, str, "path"),
size=_get(d, int, "size"),
hashes=_get_required_as(d, Mapping, _validate_hashes, "hashes"), # type: ignore[type-abstract]
)
_validate_path_url(package_wheel.path, package_wheel.url)
return package_wheel
@property
def filename(self) -> str:
"""Get the filename of the wheel."""
filename = self.name or _path_name(self.path) or _url_name(self.url)
if not filename:
raise PylockValidationError("Cannot determine wheel filename")
return filename
@dataclass(frozen=True, init=False)
class Package:
name: NormalizedName
version: Version | None = None
marker: Marker | None = None
requires_python: SpecifierSet | None = None
dependencies: Sequence[Mapping[str, Any]] | None = None
vcs: PackageVcs | None = None
directory: PackageDirectory | None = None
archive: PackageArchive | None = None
index: str | None = None
sdist: PackageSdist | None = None
wheels: Sequence[PackageWheel] | None = None
attestation_identities: Sequence[Mapping[str, Any]] | None = None
tool: Mapping[str, Any] | None = None
def __init__(
self,
*,
name: NormalizedName,
version: Version | None = None,
marker: Marker | None = None,
requires_python: SpecifierSet | None = None,
dependencies: Sequence[Mapping[str, Any]] | None = None,
vcs: PackageVcs | None = None,
directory: PackageDirectory | None = None,
archive: PackageArchive | None = None,
index: str | None = None,
sdist: PackageSdist | None = None,
wheels: Sequence[PackageWheel] | None = None,
attestation_identities: Sequence[Mapping[str, Any]] | None = None,
tool: Mapping[str, Any] | None = None,
) -> None:
# In Python 3.10+ make dataclass kw_only=True and remove __init__
object.__setattr__(self, "name", name)
object.__setattr__(self, "version", version)
object.__setattr__(self, "marker", marker)
object.__setattr__(self, "requires_python", requires_python)
object.__setattr__(self, "dependencies", dependencies)
object.__setattr__(self, "vcs", vcs)
object.__setattr__(self, "directory", directory)
object.__setattr__(self, "archive", archive)
object.__setattr__(self, "index", index)
object.__setattr__(self, "sdist", sdist)
object.__setattr__(self, "wheels", wheels)
object.__setattr__(self, "attestation_identities", attestation_identities)
object.__setattr__(self, "tool", tool)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
package = cls(
name=_get_required_as(d, str, _validate_normalized_name, "name"),
version=_get_as(d, str, Version, "version"),
requires_python=_get_as(d, str, SpecifierSet, "requires-python"),
dependencies=_get_sequence(d, Mapping, "dependencies"), # type: ignore[type-abstract]
marker=_get_as(d, str, Marker, "marker"),
vcs=_get_object(d, PackageVcs, "vcs"),
directory=_get_object(d, PackageDirectory, "directory"),
archive=_get_object(d, PackageArchive, "archive"),
index=_get(d, str, "index"),
sdist=_get_object(d, PackageSdist, "sdist"),
wheels=_get_sequence_of_objects(d, PackageWheel, "wheels"),
attestation_identities=_get_sequence(d, Mapping, "attestation-identities"), # type: ignore[type-abstract]
tool=_get(d, Mapping, "tool"), # type: ignore[type-abstract]
)
distributions = bool(package.sdist) + len(package.wheels or [])
direct_urls = (
bool(package.vcs) + bool(package.directory) + bool(package.archive)
)
if distributions > 0 and direct_urls > 0:
raise PylockValidationError(
"None of vcs, directory, archive must be set if sdist or wheels are set"
)
if distributions == 0 and direct_urls != 1:
raise PylockValidationError(
"Exactly one of vcs, directory, archive must be set "
"if sdist and wheels are not set"
)
for i, wheel in enumerate(package.wheels or []):
try:
(name, version, _, _) = parse_wheel_filename(wheel.filename)
except Exception as e:
raise PylockValidationError(
f"Invalid wheel filename {wheel.filename!r}",
context=f"wheels[{i}]",
) from e
if name != package.name:
raise PylockValidationError(
f"Name in {wheel.filename!r} is not consistent with "
f"package name {package.name!r}",
context=f"wheels[{i}]",
)
if package.version and version != package.version:
raise PylockValidationError(
f"Version in {wheel.filename!r} is not consistent with "
f"package version {str(package.version)!r}",
context=f"wheels[{i}]",
)
if package.sdist:
try:
name, version = parse_sdist_filename(package.sdist.filename)
except Exception as e:
raise PylockValidationError(
f"Invalid sdist filename {package.sdist.filename!r}",
context="sdist",
) from e
if name != package.name:
raise PylockValidationError(
f"Name in {package.sdist.filename!r} is not consistent with "
f"package name {package.name!r}",
context="sdist",
)
if package.version and version != package.version:
raise PylockValidationError(
f"Version in {package.sdist.filename!r} is not consistent with "
f"package version {str(package.version)!r}",
context="sdist",
)
try:
for i, attestation_identity in enumerate( # noqa: B007
package.attestation_identities or []
):
_get_required(attestation_identity, str, "kind")
except Exception as e:
raise PylockValidationError(
e, context=f"attestation-identities[{i}]"
) from e
return package
@property
def is_direct(self) -> bool:
return not (self.sdist or self.wheels)
@dataclass(frozen=True, init=False)
class Pylock:
"""A class representing a pylock file."""
lock_version: Version
environments: Sequence[Marker] | None = None
requires_python: SpecifierSet | None = None
extras: Sequence[NormalizedName] | None = None
dependency_groups: Sequence[str] | None = None
default_groups: Sequence[str] | None = None
created_by: str # type: ignore[misc]
packages: Sequence[Package] # type: ignore[misc]
tool: Mapping[str, Any] | None = None
def __init__(
self,
*,
lock_version: Version,
environments: Sequence[Marker] | None = None,
requires_python: SpecifierSet | None = None,
extras: Sequence[NormalizedName] | None = None,
dependency_groups: Sequence[str] | None = None,
default_groups: Sequence[str] | None = None,
created_by: str,
packages: Sequence[Package],
tool: Mapping[str, Any] | None = None,
) -> None:
# In Python 3.10+ make dataclass kw_only=True and remove __init__
object.__setattr__(self, "lock_version", lock_version)
object.__setattr__(self, "environments", environments)
object.__setattr__(self, "requires_python", requires_python)
object.__setattr__(self, "extras", extras)
object.__setattr__(self, "dependency_groups", dependency_groups)
object.__setattr__(self, "default_groups", default_groups)
object.__setattr__(self, "created_by", created_by)
object.__setattr__(self, "packages", packages)
object.__setattr__(self, "tool", tool)
@classmethod
def _from_dict(cls, d: Mapping[str, Any]) -> Self:
pylock = cls(
lock_version=_get_required_as(d, str, Version, "lock-version"),
environments=_get_sequence_as(d, str, Marker, "environments"),
extras=_get_sequence_as(d, str, _validate_normalized_name, "extras"),
dependency_groups=_get_sequence(d, str, "dependency-groups"),
default_groups=_get_sequence(d, str, "default-groups"),
created_by=_get_required(d, str, "created-by"),
requires_python=_get_as(d, str, SpecifierSet, "requires-python"),
packages=_get_required_sequence_of_objects(d, Package, "packages"),
tool=_get(d, Mapping, "tool"), # type: ignore[type-abstract]
)
if not Version("1") <= pylock.lock_version < Version("2"):
raise PylockUnsupportedVersionError(
f"pylock version {pylock.lock_version} is not supported"
)
if pylock.lock_version > Version("1.0"):
_logger.warning(
"pylock minor version %s is not supported", pylock.lock_version
)
return pylock
@classmethod
def from_dict(cls, d: Mapping[str, Any], /) -> Self:
"""Create and validate a Pylock instance from a TOML dictionary.
Raises :class:`PylockValidationError` if the input data is not
spec-compliant.
"""
return cls._from_dict(d)
def to_dict(self) -> Mapping[str, Any]:
"""Convert the Pylock instance to a TOML dictionary."""
return dataclasses.asdict(self, dict_factory=_toml_dict_factory)
def validate(self) -> None:
"""Validate the Pylock instance against the specification.
Raises :class:`PylockValidationError` otherwise."""
self.from_dict(self.to_dict())
def select(
self,
*,
environment: Environment | None = None,
tags: Sequence[Tag] | None = None,
extras: Collection[str] | None = None,
dependency_groups: Collection[str] | None = None,
) -> Iterator[
tuple[
Package,
PackageVcs
| PackageDirectory
| PackageArchive
| PackageWheel
| PackageSdist,
]
]:
"""Select what to install from the lock file.
The *environment* and *tags* parameters represent the environment being
selected for. If unspecified, ``packaging.markers.default_environment()`` and
``packaging.tags.sys_tags()`` are used.
The *extras* parameter represents the extras to install.
The *dependency_groups* parameter represents the groups to install. If
unspecified, the default groups are used.
This method must be used on valid Pylock instances (i.e. one obtained
from :meth:`Pylock.from_dict` or if constructed manually, after calling
:meth:`Pylock.validate`).
"""
compatible_tags_selector = create_compatible_tags_selector(tags or sys_tags())
# #. Gather the extras and dependency groups to install and set ``extras`` and
# ``dependency_groups`` for marker evaluation, respectively.
#
# #. ``extras`` SHOULD be set to the empty set by default.
# #. ``dependency_groups`` SHOULD be the set created from
# :ref:`pylock-default-groups` by default.
env = cast(
"dict[str, str | frozenset[str]]",
dict(
environment or {}, # Marker.evaluate will fill-up
extras=frozenset(extras or []),
dependency_groups=frozenset(
(self.default_groups or [])
if dependency_groups is None # to allow selecting no group
else dependency_groups
),
),
)
env_python_full_version = (
environment["python_full_version"]
if environment
else default_environment()["python_full_version"]
)
# #. Check if the metadata version specified by :ref:`pylock-lock-version` is
# supported; an error or warning MUST be raised as appropriate.
# Covered by lock.validate() which is a precondition for this method.
# #. If :ref:`pylock-requires-python` is specified, check that the environment
# being installed for meets the requirement; an error MUST be raised if it is
# not met.
if self.requires_python and not self.requires_python.contains(
env_python_full_version,
):
raise PylockSelectError(
f"python_full_version {env_python_full_version!r} "
f"in provided environment does not satisfy the Python version "
f"requirement {str(self.requires_python)!r}"
)
# #. If :ref:`pylock-environments` is specified, check that at least one of the
# environment marker expressions is satisfied; an error MUST be raised if no
# expression is satisfied.
if self.environments:
for env_marker in self.environments:
if env_marker.evaluate(
cast("dict[str, str]", environment or {}), context="requirement"
):
break
else:
raise PylockSelectError(
"Provided environment does not satisfy any of the "
"environments specified in the lock file"
)
# #. For each package listed in :ref:`pylock-packages`:
selected_packages_by_name: dict[str, tuple[int, Package]] = {}
for package_index, package in enumerate(self.packages):
# #. If :ref:`pylock-packages-marker` is specified, check if it is
# satisfied;if it isn't, skip to the next package.
if package.marker and not package.marker.evaluate(env, context="lock_file"):
continue
# #. If :ref:`pylock-packages-requires-python` is specified, check if it is
# satisfied; an error MUST be raised if it isn't.
if package.requires_python and not package.requires_python.contains(
env_python_full_version,
):
raise PylockSelectError(
f"python_full_version {env_python_full_version!r} "
f"in provided environment does not satisfy the Python version "
f"requirement {str(package.requires_python)!r} for package "
f"{package.name!r} at packages[{package_index}]"
)
# #. Check that no other conflicting instance of the package has been slated
# to be installed; an error about the ambiguity MUST be raised otherwise.
if package.name in selected_packages_by_name:
raise PylockSelectError(
f"Multiple packages with the name {package.name!r} are "
f"selected at packages[{package_index}] and "
f"packages[{selected_packages_by_name[package.name][0]}]"
)
# #. Check that the source of the package is specified appropriately (i.e.
# there are no conflicting sources in the package entry);
# an error MUST be raised if any issues are found.
# Covered by lock.validate() which is a precondition for this method.
# #. Add the package to the set of packages to install.
selected_packages_by_name[package.name] = (package_index, package)
# #. For each package to be installed:
for package_index, package in selected_packages_by_name.values():
# - If :ref:`pylock-packages-vcs` is set:
if package.vcs is not None:
yield package, package.vcs
# - Else if :ref:`pylock-packages-directory` is set:
elif package.directory is not None:
yield package, package.directory
# - Else if :ref:`pylock-packages-archive` is set:
elif package.archive is not None:
yield package, package.archive
# - Else if there are entries for :ref:`pylock-packages-wheels`:
elif package.wheels:
# #. Look for the appropriate wheel file based on
# :ref:`pylock-packages-wheels-name`; if one is not found then move
# on to :ref:`pylock-packages-sdist` or an error MUST be raised about
# a lack of source for the project.
best_wheel = next(
compatible_tags_selector(
(wheel, parse_wheel_filename(wheel.filename)[-1])
for wheel in package.wheels
),
None,
)
if best_wheel:
yield package, best_wheel
elif package.sdist is not None:
yield package, package.sdist
else:
raise PylockSelectError(
f"No wheel found matching the provided tags "
f"for package {package.name!r} "
f"at packages[{package_index}], "
f"and no sdist available as a fallback"
)
# - Else if no :ref:`pylock-packages-wheels` file is found or
# :ref:`pylock-packages-sdist` is solely set:
elif package.sdist is not None:
yield package, package.sdist
else:
# Covered by lock.validate() which is a precondition for this method.
raise NotImplementedError # pragma: no cover
+48 -10
View File
@@ -3,7 +3,7 @@
# for complete details.
from __future__ import annotations
from typing import Any, Iterator
from typing import Iterator
from ._parser import parse_requirement as _parse_requirement
from ._tokenizer import ParserSyntaxError
@@ -11,6 +11,15 @@ from .markers import Marker, _normalize_extra_values
from .specifiers import SpecifierSet
from .utils import canonicalize_name
__all__ = [
"InvalidRequirement",
"Requirement",
]
def __dir__() -> list[str]:
return __all__
class InvalidRequirement(ValueError):
"""
@@ -24,6 +33,16 @@ class Requirement:
Parse a given requirement string into its parts, such as name, specifier,
URL, and extras. Raises InvalidRequirement on a badly-formed requirement
string.
Instances are safe to serialize with :mod:`pickle`. They use a stable
format so the same pickle can be loaded in future packaging releases.
.. versionchanged:: 26.2
Added a stable pickle format. Pickles created with packaging 26.2+ can
be unpickled with future releases. Backward compatibility with pickles
from packaging < 26.2 is supported but may be removed in a future
release.
"""
# TODO: Can we test whether something is contained within a requirement?
@@ -57,28 +76,47 @@ class Requirement:
yield str(self.specifier)
if self.url:
yield f"@ {self.url}"
yield f" @ {self.url}"
if self.marker:
yield " "
if self.marker:
yield f"; {self.marker}"
def __getstate__(self) -> str:
# Return the requirement string for compactness and stability.
# Re-parsed on load to reconstruct all fields.
return str(self)
def __setstate__(self, state: object) -> None:
if isinstance(state, str):
# New format (26.2+): just the requirement string.
try:
tmp = Requirement(state)
except InvalidRequirement as exc:
raise TypeError(f"Cannot restore Requirement from {state!r}") from exc
self.name = tmp.name
self.url = tmp.url
self.extras = tmp.extras
self.specifier = tmp.specifier
self.marker = tmp.marker
return
if isinstance(state, dict):
# Old format (packaging <= 26.1, no __slots__): plain __dict__.
self.__dict__.update(state)
return
raise TypeError(f"Cannot restore Requirement from {state!r}")
def __str__(self) -> str:
return "".join(self._iter_parts(self.name))
def __repr__(self) -> str:
return f"<Requirement('{self}')>"
return f"<{self.__class__.__name__}({str(self)!r})>"
def __hash__(self) -> int:
return hash(
(
self.__class__.__name__,
*self._iter_parts(canonicalize_name(self.name)),
)
)
return hash(tuple(self._iter_parts(canonicalize_name(self.name))))
def __eq__(self, other: Any) -> bool:
def __eq__(self, other: object) -> bool:
if not isinstance(other, Requirement):
return NotImplemented
File diff suppressed because it is too large Load Diff
+337 -61
View File
@@ -5,6 +5,7 @@
from __future__ import annotations
import logging
import operator
import platform
import re
import struct
@@ -13,19 +14,52 @@ import sys
import sysconfig
from importlib.machinery import EXTENSION_SUFFIXES
from typing import (
TYPE_CHECKING,
Iterable,
Iterator,
Sequence,
Tuple,
TypeVar,
cast,
)
from . import _manylinux, _musllinux
if TYPE_CHECKING:
from collections.abc import Callable, Iterable
from typing import AbstractSet
__all__ = [
"INTERPRETER_SHORT_NAMES",
"AppleVersion",
"PythonVersion",
"Tag",
"UnsortedTagsError",
"android_platforms",
"compatible_tags",
"cpython_tags",
"create_compatible_tags_selector",
"generic_tags",
"interpreter_name",
"interpreter_version",
"ios_platforms",
"mac_platforms",
"parse_tag",
"platform_tags",
"sys_tags",
]
def __dir__() -> list[str]:
return __all__
logger = logging.getLogger(__name__)
PythonVersion = Sequence[int]
AppleVersion = Tuple[int, int]
_T = TypeVar("_T")
INTERPRETER_SHORT_NAMES: dict[str, str] = {
"python": "py", # Generic.
@@ -36,7 +70,19 @@ INTERPRETER_SHORT_NAMES: dict[str, str] = {
}
_32_BIT_INTERPRETER = struct.calcsize("P") == 4
# This function can be unit tested without reloading the module
# (Unlike _32_BIT_INTERPRETER)
def _compute_32_bit_interpreter() -> bool:
return struct.calcsize("P") == 4
_32_BIT_INTERPRETER = _compute_32_bit_interpreter()
class UnsortedTagsError(ValueError):
"""
Raised when a tag component is not in sorted order per PEP 425.
"""
class Tag:
@@ -45,11 +91,29 @@ class Tag:
Instances are considered immutable and thus are hashable. Equality checking
is also supported.
Instances are safe to serialize with :mod:`pickle`. They use a stable
format so the same pickle can be loaded in future packaging releases.
.. versionchanged:: 26.2
Added a stable pickle format. Pickles created with packaging 26.2+ can
be unpickled with future releases. Backward compatibility with pickles
from packaging < 26.2 is supported but may be removed in a future
release.
"""
__slots__ = ["_abi", "_hash", "_interpreter", "_platform"]
def __init__(self, interpreter: str, abi: str, platform: str) -> None:
"""
:param str interpreter: The interpreter name, e.g. ``"py"``
(see :attr:`INTERPRETER_SHORT_NAMES` for mapping
well-known interpreter names to their short names).
:param str abi: The ABI that a wheel supports, e.g. ``"cp37m"``.
:param str platform: The OS/platform the wheel supports,
e.g. ``"win_amd64"``.
"""
self._interpreter = interpreter.lower()
self._abi = abi.lower()
self._platform = platform.lower()
@@ -62,14 +126,25 @@ class Tag:
@property
def interpreter(self) -> str:
"""
The interpreter name, e.g. ``"py"`` (see
:attr:`INTERPRETER_SHORT_NAMES` for mapping well-known interpreter
names to their short names).
"""
return self._interpreter
@property
def abi(self) -> str:
"""
The supported ABI.
"""
return self._abi
@property
def platform(self) -> str:
"""
The OS/platform.
"""
return self._platform
def __eq__(self, other: object) -> bool:
@@ -92,16 +167,69 @@ class Tag:
def __repr__(self) -> str:
return f"<{self} @ {id(self)}>"
def __getstate__(self) -> tuple[str, str, str]:
# Return state as a 3-item tuple: (interpreter, abi, platform).
# Cache member _hash is excluded and will be recomputed.
return (self._interpreter, self._abi, self._platform)
def parse_tag(tag: str) -> frozenset[Tag]:
def __setstate__(self, state: object) -> None:
if isinstance(state, tuple):
if len(state) == 3 and all(isinstance(s, str) for s in state):
# New format (26.2+): (interpreter, abi, platform)
self._interpreter, self._abi, self._platform = state
self._hash = hash((self._interpreter, self._abi, self._platform))
return
if len(state) == 2 and isinstance(state[1], dict):
# Old format (packaging <= 26.1, __slots__): (None, {slot: value}).
_, slots = state
try:
interpreter = slots["_interpreter"]
abi = slots["_abi"]
platform = slots["_platform"]
except KeyError:
raise TypeError(f"Cannot restore Tag from {state!r}") from None
if not all(
isinstance(value, str) for value in (interpreter, abi, platform)
):
raise TypeError(f"Cannot restore Tag from {state!r}")
self._interpreter = interpreter.lower()
self._abi = abi.lower()
self._platform = platform.lower()
self._hash = hash((self._interpreter, self._abi, self._platform))
return
raise TypeError(f"Cannot restore Tag from {state!r}")
def parse_tag(tag: str, *, validate_order: bool = False) -> frozenset[Tag]:
"""
Parses the provided tag (e.g. `py3-none-any`) into a frozenset of Tag instances.
Parses the provided tag (e.g. `py3-none-any`) into a frozenset of
:class:`Tag` instances.
Returning a set is required due to the possibility that the tag is a
compressed tag set.
`compressed tag set`_, e.g. ``"py2.py3-none-any"`` which supports both
Python 2 and Python 3.
If **validate_order** is true, compressed tag set components are checked
to be in sorted order as required by PEP 425.
:param str tag: The tag to parse, e.g. ``"py3-none-any"``.
:param bool validate_order: Check whether compressed tag set components
are in sorted order.
:raises UnsortedTagsError: If **validate_order** is true and any compressed tag
set component is not in sorted order.
.. versionadded:: 26.1
The *validate_order* parameter.
"""
tags = set()
interpreters, abis, platforms = tag.split("-")
if validate_order:
for component in (interpreters, abis, platforms):
parts = component.split(".")
if parts != sorted(parts):
raise UnsortedTagsError(
f"Tag component {component!r} is not in sorted order per PEP 425"
)
for interpreter in interpreters.split("."):
for abi in abis.split("."):
for platform_ in platforms.split("."):
@@ -142,12 +270,25 @@ def _abi3_applies(python_version: PythonVersion, threading: bool) -> bool:
"""
Determine if the Python version supports abi3.
PEP 384 was first implemented in Python 3.2. The threaded (`--disable-gil`)
PEP 384 was first implemented in Python 3.2. The free-threaded
builds do not support abi3.
"""
return len(python_version) > 1 and tuple(python_version) >= (3, 2) and not threading
def _abi3t_applies(python_version: PythonVersion, threading: bool) -> bool:
"""
Determine if the Python version supports abi3t.
PEP 803 was first implemented in Python 3.15 but, per PEP 803, this
returns tags going back to Python 3.2 to mirror the abi3
implementation and leave open the possibility of abi3t wheels
supporting older Python versions.
"""
return len(python_version) > 1 and tuple(python_version) >= (3, 2) and threading
def _cpython_abis(py_version: PythonVersion, warn: bool = False) -> list[str]:
py_version = tuple(py_version) # To allow for version comparison.
abis = []
@@ -189,19 +330,31 @@ def cpython_tags(
warn: bool = False,
) -> Iterator[Tag]:
"""
Yields the tags for a CPython interpreter.
Yields the tags for the CPython interpreter.
The tags consist of:
- cp<python_version>-<abi>-<platform>
- cp<python_version>-abi3-<platform>
- cp<python_version>-none-<platform>
- cp<less than python_version>-abi3-<platform> # Older Python versions down to 3.2.
The specific tags generated are:
If python_version only specifies a major version then user-provided ABIs and
the 'none' ABItag will be used.
- ``cp<python_version>-<abi>-<platform>``
- ``cp<python_version>-<stable_abi>-<platform>``
- ``cp<python_version>-none-<platform>``
- ``cp<older version>-<stable_abi>-<platform>`` where "older version" is all older
minor versions down to Python 3.2 (when ``abi3`` was introduced)
If 'abi3' or 'none' are specified in 'abis' then they will be yielded at
their normal position and not at the beginning.
If ``python_version`` only provides a major-only version then only
user-provided ABIs via ``abis`` and the ``none`` ABI will be used.
The ``stable_abi`` will be either ``abi3`` or ``abi3t`` if `abi` is a
GIL-enabled ABI like `"cp315"` or a free-threaded ABI like `"cp315t"`,
respectively.
:param Sequence python_version: A one- or two-item sequence representing the
targeted Python version. Defaults to
``sys.version_info[:2]``.
:param Iterable abis: Iterable of compatible ABIs. Defaults to the ABIs
compatible with the current system.
:param Iterable platforms: Iterable of compatible platforms. Defaults to the
platforms compatible with the current system.
:param bool warn: Whether warnings should be logged. Defaults to ``False``.
"""
if not python_version:
python_version = sys.version_info[:2]
@@ -209,16 +362,13 @@ def cpython_tags(
interpreter = f"cp{_version_nodot(python_version[:2])}"
if abis is None:
if len(python_version) > 1:
abis = _cpython_abis(python_version, warn)
else:
abis = []
abis = _cpython_abis(python_version, warn) if len(python_version) > 1 else []
abis = list(abis)
# 'abi3' and 'none' are explicitly handled later.
for explicit_abi in ("abi3", "none"):
try:
abis.remove(explicit_abi)
except ValueError:
except ValueError: # noqa: PERF203
pass
platforms = list(platforms or platform_tags())
@@ -228,16 +378,27 @@ def cpython_tags(
threading = _is_threaded_cpython(abis)
use_abi3 = _abi3_applies(python_version, threading)
if use_abi3:
yield from (Tag(interpreter, "abi3", platform_) for platform_ in platforms)
yield from (Tag(interpreter, "none", platform_) for platform_ in platforms)
use_abi3t = _abi3t_applies(python_version, threading)
if use_abi3:
yield from (Tag(interpreter, "abi3", platform_) for platform_ in platforms)
if use_abi3t:
yield from (Tag(interpreter, "abi3t", platform_) for platform_ in platforms)
yield from (Tag(interpreter, "none", platform_) for platform_ in platforms)
if use_abi3 or use_abi3t:
for minor_version in range(python_version[1] - 1, 1, -1):
for platform_ in platforms:
version = _version_nodot((python_version[0], minor_version))
interpreter = f"cp{version}"
yield Tag(interpreter, "abi3", platform_)
if use_abi3:
yield Tag(interpreter, "abi3", platform_)
if use_abi3t:
# Support for abi3t was introduced in Python 3.15, but in
# principle abi3t wheels are possible for older limited API
# versions, so allow things like ("cp37", "abi3t", "platform")
yield Tag(interpreter, "abi3t", platform_)
def _generic_abi() -> list[str]:
@@ -289,21 +450,31 @@ def generic_tags(
warn: bool = False,
) -> Iterator[Tag]:
"""
Yields the tags for a generic interpreter.
Yields the tags for an interpreter which requires no specialization.
The tags consist of:
- <interpreter>-<abi>-<platform>
This function should be used if one of the other interpreter-specific
functions provided by this module is not appropriate (i.e. not calculating
tags for a CPython interpreter).
The "none" ABI will be added if it was not explicitly provided.
The specific tags generated are:
- ``<interpreter>-<abi>-<platform>``
The ``"none"`` ABI will be added if it was not explicitly provided.
:param str interpreter: The name of the interpreter. Defaults to being
calculated.
:param Iterable abis: Iterable of compatible ABIs. Defaults to the ABIs
compatible with the current system.
:param Iterable platforms: Iterable of compatible platforms. Defaults to the
platforms compatible with the current system.
:param bool warn: Whether warnings should be logged. Defaults to ``False``.
"""
if not interpreter:
interp_name = interpreter_name()
interp_version = interpreter_version(warn=warn)
interpreter = "".join([interp_name, interp_version])
if abis is None:
abis = _generic_abi()
else:
abis = list(abis)
interpreter = f"{interp_name}{interp_version}"
abis = _generic_abi() if abis is None else list(abis)
platforms = list(platforms or platform_tags())
if "none" not in abis:
abis.append("none")
@@ -333,12 +504,22 @@ def compatible_tags(
platforms: Iterable[str] | None = None,
) -> Iterator[Tag]:
"""
Yields the sequence of tags that are compatible with a specific version of Python.
Yields the tags for an interpreter compatible with the Python version
specified by ``python_version``.
The tags consist of:
- py*-none-<platform>
- <interpreter>-none-any # ... if `interpreter` is provided.
- py*-none-any
The specific tags generated are:
- ``py*-none-<platform>``
- ``<interpreter>-none-any`` if ``interpreter`` is provided
- ``py*-none-any``
:param Sequence python_version: A one- or two-item sequence representing the
compatible version of Python. Defaults to
``sys.version_info[:2]``.
:param str interpreter: The name of the interpreter (if known), e.g.
``"cp38"``. Defaults to the current interpreter.
:param Iterable platforms: Iterable of compatible platforms. Defaults to the
platforms compatible with the current system.
"""
if not python_version:
python_version = sys.version_info[:2]
@@ -398,12 +579,25 @@ def mac_platforms(
version: AppleVersion | None = None, arch: str | None = None
) -> Iterator[str]:
"""
Yields the platform tags for a macOS system.
Yields the :attr:`~Tag.platform` tags for macOS.
The `version` parameter is a two-item tuple specifying the macOS version to
generate platform tags for. The `arch` parameter is the CPU architecture to
generate platform tags for. Both parameters default to the appropriate value
for the current system.
:param tuple version: A two-item tuple representing the version of macOS.
Defaults to the current system's version.
:param str arch: The CPU architecture. Defaults to the architecture of the
current system, e.g. ``"x86_64"``.
.. note::
Equivalent support for the other major platforms is purposefully not
provided:
- On Windows, platform compatibility is statically specified
- On Linux, code must be run on the system itself to determine
compatibility
"""
version_str, _, cpu_arch = platform.mac_ver()
if version is None:
@@ -424,14 +618,11 @@ def mac_platforms(
text=True,
).stdout
version = cast("AppleVersion", tuple(map(int, version_str.split(".")[:2])))
else:
version = version
if arch is None:
arch = _mac_arch(cpu_arch)
else:
arch = arch
if (10, 0) <= version and version < (11, 0):
if (10, 0) <= version < (11, 0):
# Prior to Mac OS 11, each yearly release of Mac OS bumped the
# "minor" version number. The major version was always 10.
major_version = 10
@@ -477,14 +668,19 @@ def ios_platforms(
version: AppleVersion | None = None, multiarch: str | None = None
) -> Iterator[str]:
"""
Yields the platform tags for an iOS system.
:param version: A two-item tuple specifying the iOS version to generate
platform tags for. Defaults to the current iOS version.
:param multiarch: The CPU architecture+ABI to generate platform tags for -
(the value used by `sys.implementation._multiarch` e.g.,
`arm64_iphoneos` or `x84_64_iphonesimulator`). Defaults to the current
multiarch value.
Yields the :attr:`~Tag.platform` tags for iOS.
:param tuple version: A two-item tuple representing the version of iOS.
Defaults to the current system's version.
:param str multiarch: The CPU architecture+ABI to be used. This should be in
the format by ``sys.implementation._multiarch`` (e.g.,
``arm64_iphoneos`` or ``x86_64_iphonesimulator``).
Defaults to the current system's multiarch value.
.. note::
Behavior of this method is undefined if invoked on non-iOS platforms
without providing explicit version and multiarch arguments.
"""
if version is None:
# if iOS is the current platform, ios_ver *must* be defined. However,
@@ -586,13 +782,22 @@ def _linux_platforms(is_32bit: bool = _32_BIT_INTERPRETER) -> Iterator[str]:
yield f"linux_{arch}"
def _emscripten_platforms() -> Iterator[str]:
pyemscripten_platform_version = sysconfig.get_config_var(
"PYEMSCRIPTEN_PLATFORM_VERSION"
)
if pyemscripten_platform_version:
yield f"pyemscripten_{pyemscripten_platform_version}_wasm32"
yield from _generic_platforms()
def _generic_platforms() -> Iterator[str]:
yield _normalize_string(sysconfig.get_platform())
def platform_tags() -> Iterator[str]:
"""
Provides the platform tags for this installation.
Yields the :attr:`~Tag.platform` tags for the running interpreter.
"""
if platform.system() == "Darwin":
return mac_platforms()
@@ -602,6 +807,8 @@ def platform_tags() -> Iterator[str]:
return android_platforms()
elif platform.system() == "Linux":
return _linux_platforms()
elif platform.system() == "Emscripten":
return _emscripten_platforms()
else:
return _generic_platforms()
@@ -612,6 +819,8 @@ def interpreter_name() -> str:
Some implementations have a reserved, two-letter abbreviation which will
be returned when appropriate.
This typically acts as the prefix to the :attr:`~Tag.interpreter` tag.
"""
name = sys.implementation.name
return INTERPRETER_SHORT_NAMES.get(name) or name
@@ -619,14 +828,14 @@ def interpreter_name() -> str:
def interpreter_version(*, warn: bool = False) -> str:
"""
Returns the version of the running interpreter.
Returns the running interpreter's version.
This typically acts as the suffix to the :attr:`~Tag.interpreter` tag.
:param bool warn: Whether warnings should be logged. Defaults to ``False``.
"""
version = _get_config_var("py_version_nodot", warn=warn)
if version:
version = str(version)
else:
version = _version_nodot(sys.version_info[:2])
return version
return str(version) if version else _version_nodot(sys.version_info[:2])
def _version_nodot(version: PythonVersion) -> str:
@@ -635,10 +844,31 @@ def _version_nodot(version: PythonVersion) -> str:
def sys_tags(*, warn: bool = False) -> Iterator[Tag]:
"""
Returns the sequence of tag triples for the running interpreter.
Yields the sequence of tag triples that the running interpreter supports.
The order of the sequence corresponds to priority order for the
interpreter, from most to least important.
The iterable is ordered so that the best-matching tag is first in the
sequence. The exact preferential order to tags is interpreter-specific, but
in general the tag importance is in the order of:
1. Interpreter
2. Platform
3. ABI
This order is due to the fact that an ABI is inherently tied to the
platform, but platform-specific code is not necessarily tied to the ABI. The
interpreter is the most important tag as it dictates basic support for any
wheel.
The function returns an iterable in order to allow for the possible
short-circuiting of tag generation if the entire sequence is not necessary
and tag calculation happens to be expensive.
:param bool warn: Whether warnings should be logged. Defaults to ``False``.
.. versionchanged:: 21.3
Added the `pp3-none-any` tag (:issue:`311`).
.. versionchanged:: 27.0
Added the `abi3t` tag (:issue:`1099`).
"""
interp_name = interpreter_name()
@@ -654,3 +884,49 @@ def sys_tags(*, warn: bool = False) -> Iterator[Tag]:
else:
interp = None
yield from compatible_tags(interpreter=interp)
def create_compatible_tags_selector(
tags: Iterable[Tag],
) -> Callable[[Iterable[tuple[_T, AbstractSet[Tag]]]], Iterator[_T]]:
"""Create a callable to select things compatible with supported tags.
This function accepts an ordered sequence of tags, with the preferred
tags first.
The returned callable accepts an iterable of tuples (thing, set[Tag]),
and returns an iterator of things, with the things with the best
matching tags first.
Example to select compatible wheel filenames:
>>> from packaging import tags
>>> from packaging.utils import parse_wheel_filename
>>> selector = tags.create_compatible_tags_selector(tags.sys_tags())
>>> filenames = ["foo-1.0-py3-none-any.whl", "foo-1.0-py2-none-any.whl"]
>>> list(selector([
... (filename, parse_wheel_filename(filename)[-1]) for filename in filenames
... ]))
['foo-1.0-py3-none-any.whl']
.. versionadded:: 26.1
"""
tag_ranks: dict[Tag, int] = {}
for rank, tag in enumerate(tags):
tag_ranks.setdefault(tag, rank) # ignore duplicate tags, keep first
supported_tags = tag_ranks.keys()
def selector(
tagged_things: Iterable[tuple[_T, AbstractSet[Tag]]],
) -> Iterator[_T]:
ranked_things: list[tuple[_T, int]] = []
for thing, thing_tags in tagged_things:
supported_thing_tags = thing_tags & supported_tags
if supported_thing_tags:
thing_rank = min(tag_ranks[t] for t in supported_thing_tags)
ranked_things.append((thing, thing_rank))
return iter(
thing for thing, _ in sorted(ranked_things, key=operator.itemgetter(1))
)
return selector
+160 -27
View File
@@ -4,15 +4,36 @@
from __future__ import annotations
import functools
import re
from typing import NewType, Tuple, Union, cast
from .tags import Tag, parse_tag
from .tags import Tag, UnsortedTagsError, parse_tag
from .version import InvalidVersion, Version, _TrimmedRelease
__all__ = [
"BuildTag",
"InvalidName",
"InvalidSdistFilename",
"InvalidWheelFilename",
"NormalizedName",
"canonicalize_name",
"canonicalize_version",
"is_normalized_name",
"parse_sdist_filename",
"parse_wheel_filename",
]
def __dir__() -> list[str]:
return __all__
BuildTag = Union[Tuple[()], Tuple[int, str]]
NormalizedName = NewType("NormalizedName", str)
"""
A :class:`typing.NewType` of :class:`str`, representing a normalized name.
"""
class InvalidName(ValueError):
@@ -35,33 +56,77 @@ class InvalidSdistFilename(ValueError):
# Core metadata spec for `Name`
_validate_regex = re.compile(
r"^([A-Z0-9]|[A-Z0-9][A-Z0-9._-]*[A-Z0-9])$", re.IGNORECASE
r"[a-z0-9]|[a-z0-9][a-z0-9._-]*[a-z0-9]", re.IGNORECASE | re.ASCII
)
_canonicalize_regex = re.compile(r"[-_.]+")
_normalized_regex = re.compile(r"^([a-z0-9]|[a-z0-9]([a-z0-9-](?!--))*[a-z0-9])$")
_normalized_regex = re.compile(r"[a-z0-9]|[a-z0-9]([a-z0-9-](?!--))*[a-z0-9]", re.ASCII)
# PEP 427: The build number must start with a digit.
_build_tag_regex = re.compile(r"(\d+)(.*)")
_build_tag_regex = re.compile(r"(\d+)(.*)", re.ASCII)
def canonicalize_name(name: str, *, validate: bool = False) -> NormalizedName:
if validate and not _validate_regex.match(name):
"""
This function takes a valid Python package or extra name, and returns the
normalized form of it.
The return type is typed as :class:`NormalizedName`. This allows type
checkers to help require that a string has passed through this function
before use.
If **validate** is true, then the function will check if **name** is a valid
distribution name before normalizing.
:param str name: The name to normalize.
:param bool validate: Check whether the name is a valid distribution name.
:raises InvalidName: If **validate** is true and the name is not an
acceptable distribution name.
>>> from packaging.utils import canonicalize_name
>>> canonicalize_name("Django")
'django'
>>> canonicalize_name("oslo.concurrency")
'oslo-concurrency'
>>> canonicalize_name("requests")
'requests'
"""
if validate and not _validate_regex.fullmatch(name):
raise InvalidName(f"name is invalid: {name!r}")
# This is taken from PEP 503.
value = _canonicalize_regex.sub("-", name).lower()
return cast(NormalizedName, value)
# Ensure all ``.`` and ``_`` are ``-``
# Emulates ``re.sub(r"[-_.]+", "-", name).lower()`` from PEP 503
# Much faster than re, and even faster than str.translate
value = name.lower().replace("_", "-").replace(".", "-")
# Condense repeats (faster than regex)
while "--" in value:
value = value.replace("--", "-")
return cast("NormalizedName", value)
def is_normalized_name(name: str) -> bool:
return _normalized_regex.match(name) is not None
"""
Check if a name is already normalized (i.e. :func:`canonicalize_name` would
roundtrip to the same value).
:param str name: The name to check.
>>> from packaging.utils import is_normalized_name
>>> is_normalized_name("requests")
True
>>> is_normalized_name("Django")
False
"""
return _normalized_regex.fullmatch(name) is not None
@functools.singledispatch
def canonicalize_version(
version: Version | str, *, strip_trailing_zero: bool = True
) -> str:
"""
Return a canonical form of a version as a string.
"""Return a canonical form of a version as a string.
This function takes a string representing a package version (or a
:class:`~packaging.version.Version` instance), and returns the
normalized form of it. By default, it strips trailing zeros from
the release segment.
>>> from packaging.utils import canonicalize_version
>>> canonicalize_version('1.0.1')
'1.0.1'
@@ -77,23 +142,63 @@ def canonicalize_version(
>>> canonicalize_version('foo bar baz')
'foo bar baz'
>>> canonicalize_version('1.4.0.0.0')
'1.4'
"""
return str(_TrimmedRelease(str(version)) if strip_trailing_zero else version)
@canonicalize_version.register
def _(version: str, *, strip_trailing_zero: bool = True) -> str:
try:
parsed = Version(version)
except InvalidVersion:
# Legacy versions cannot be normalized
return version
return canonicalize_version(parsed, strip_trailing_zero=strip_trailing_zero)
if isinstance(version, str):
try:
version = Version(version)
except InvalidVersion:
return str(version)
return str(_TrimmedRelease(version) if strip_trailing_zero else version)
def parse_wheel_filename(
filename: str,
*,
validate_order: bool = False,
) -> tuple[NormalizedName, Version, BuildTag, frozenset[Tag]]:
"""
This function takes the filename of a wheel file, and parses it,
returning a tuple of name, version, build number, and tags.
The name part of the tuple is normalized and typed as
:class:`NormalizedName`. The version portion is an instance of
:class:`~packaging.version.Version`. The build number is ``()`` if
there is no build number in the wheel filename, otherwise a
two-item tuple of an integer for the leading digits and
a string for the rest of the build number. The tags portion is a
frozen set of :class:`~packaging.tags.Tag` instances (as the tag
string format allows multiple tags to be combined into a single
string).
If **validate_order** is true, compressed tag set components are
checked to be in sorted order as required by PEP 425.
:param str filename: The name of the wheel file.
:param bool validate_order: Check whether compressed tag set components
are in sorted order.
:raises InvalidWheelFilename: If the filename in question
does not follow the :ref:`wheel specification
<pypug:binary-distribution-format>`.
>>> from packaging.utils import parse_wheel_filename
>>> from packaging.tags import Tag
>>> from packaging.version import Version
>>> name, ver, build, tags = parse_wheel_filename("foo-1.0-py3-none-any.whl")
>>> name
'foo'
>>> ver == Version('1.0')
True
>>> tags == {Tag("py3", "none", "any")}
True
>>> not build
True
.. versionadded:: 26.1
The *validate_order* parameter.
"""
if not filename.endswith(".whl"):
raise InvalidWheelFilename(
f"Invalid wheel filename (extension must be '.whl'): {filename!r}"
@@ -127,14 +232,42 @@ def parse_wheel_filename(
raise InvalidWheelFilename(
f"Invalid build number: {build_part} in {filename!r}"
)
build = cast(BuildTag, (int(build_match.group(1)), build_match.group(2)))
build = cast("BuildTag", (int(build_match.group(1)), build_match.group(2)))
else:
build = ()
tags = parse_tag(parts[-1])
tag_str = parts[-1]
try:
tags = parse_tag(tag_str, validate_order=validate_order)
except UnsortedTagsError:
raise InvalidWheelFilename(
f"Invalid wheel filename (compressed tag set components must be in "
f"sorted order per PEP 425): {filename!r}"
) from None
return (name, version, build, tags)
def parse_sdist_filename(filename: str) -> tuple[NormalizedName, Version]:
"""
This function takes the filename of a sdist file (as specified
in the `Source distribution format`_ documentation), and parses
it, returning a tuple of the normalized name and version as
represented by an instance of :class:`~packaging.version.Version`.
:param str filename: The name of the sdist file.
:raises InvalidSdistFilename: If the filename does not end
with an sdist extension (``.zip`` or ``.tar.gz``), or if it does not
contain a dash separating the name and the version of the distribution.
>>> from packaging.utils import parse_sdist_filename
>>> from packaging.version import Version
>>> name, ver = parse_sdist_filename("foo-1.0.tar.gz")
>>> name
'foo'
>>> ver == Version('1.0')
True
.. _Source distribution format: https://packaging.python.org/specifications/source-distribution-format/#source-distribution-file-name
"""
if filename.endswith(".tar.gz"):
file_stem = filename[: -len(".tar.gz")]
elif filename.endswith(".zip"):
File diff suppressed because it is too large Load Diff